Skip to content

F
face
Unverified Commit b0d514d5 by justadudewhohacks Committed by GitHub
Options

Merge pull request #41 from justadudewhohacks/disposable

param mappings for remaining nets + make face detection and recognition nets extend NeuralNetwork
parents a6a68a56 5e0a71fc
Showing with 1948 additions and 1651 deletions
build/Rect.d.ts
...@@ -11,5 +11,6 @@ export declare class Rect implements IRect { ...@@ -11,5 +11,6 @@ export declare class Rect implements IRect {
height: number; height: number;
constructor(x: number, y: number, width: number, height: number); constructor(x: number, y: number, width: number, height: number);
toSquare(): Rect; toSquare(): Rect;
pad(padX: number, padY: number): Rect;
floor(): Rect; floor(): Rect;
} }
build/Rect.js
...@@ -20,6 +20,10 @@ var Rect = /** @class */ (function () { ...@@ -20,6 +20,10 @@ var Rect = /** @class */ (function () {
} }
return new Rect(x, y, width, height); return new Rect(x, y, width, height);
}; };
Rect.prototype.pad = function (padX, padY) {
var _a = this, x = _a.x, y = _a.y, width = _a.width, height = _a.height;
return new Rect(x - (padX / 2), y - (padY / 2), width + padX, height + padY);
};
Rect.prototype.floor = function () { Rect.prototype.floor = function () {
return new Rect(Math.floor(this.x), Math.floor(this.y), Math.floor(this.width), Math.floor(this.height)); return new Rect(Math.floor(this.x), Math.floor(this.y), Math.floor(this.width), Math.floor(this.height));
}; };
......
build/Rect.js.map
{"version":3,"file":"Rect.js","sourceRoot":"","sources":["../src/Rect.ts"],"names":[],"mappings":";;AAOA;IAME,cAAY,CAAS,EAAE,CAAS,EAAE,KAAa,EAAE,MAAc;QAC7D,IAAI,CAAC,CAAC,GAAG,CAAC,CAAA;QACV,IAAI,CAAC,CAAC,GAAG,CAAC,CAAA;QACV,IAAI,CAAC,KAAK,GAAG,KAAK,CAAA;QAClB,IAAI,CAAC,MAAM,GAAG,MAAM,CAAA;IACtB,CAAC;IAEM,uBAAQ,GAAf;QACM,IAAA,SAA8B,EAA5B,QAAC,EAAE,QAAC,EAAE,gBAAK,EAAE,kBAAM,CAAS;QAClC,IAAM,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,KAAK,GAAG,MAAM,CAAC,CAAA;QACrC,IAAI,KAAK,GAAG,MAAM,EAAE;YAClB,CAAC,IAAI,CAAC,IAAI,GAAG,CAAC,CAAC,CAAA;YACf,KAAK,IAAI,IAAI,CAAA;SACd;QACD,IAAI,MAAM,GAAG,KAAK,EAAE;YAClB,CAAC,IAAI,CAAC,IAAI,GAAG,CAAC,CAAC,CAAA;YACf,MAAM,IAAI,IAAI,CAAA;SACf;QACD,OAAO,IAAI,IAAI,CAAC,CAAC,EAAE,CAAC,EAAE,KAAK,EAAE,MAAM,CAAC,CAAA;IACtC,CAAC;IAEM,oBAAK,GAAZ;QACE,OAAO,IAAI,IAAI,CACb,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,EAClB,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,EAClB,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,EACtB,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,CACxB,CAAA;IACH,CAAC;IACH,WAAC;AAAD,CAAC,AAnCD,IAmCC;AAnCY,oBAAI"} {"version":3,"file":"Rect.js","sourceRoot":"","sources":["../src/Rect.ts"],"names":[],"mappings":";;AAOA;IAME,cAAY,CAAS,EAAE,CAAS,EAAE,KAAa,EAAE,MAAc;QAC7D,IAAI,CAAC,CAAC,GAAG,CAAC,CAAA;QACV,IAAI,CAAC,CAAC,GAAG,CAAC,CAAA;QACV,IAAI,CAAC,KAAK,GAAG,KAAK,CAAA;QAClB,IAAI,CAAC,MAAM,GAAG,MAAM,CAAA;IACtB,CAAC;IAEM,uBAAQ,GAAf;QACM,IAAA,SAA8B,EAA5B,QAAC,EAAE,QAAC,EAAE,gBAAK,EAAE,kBAAM,CAAS;QAClC,IAAM,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,KAAK,GAAG,MAAM,CAAC,CAAA;QACrC,IAAI,KAAK,GAAG,MAAM,EAAE;YAClB,CAAC,IAAI,CAAC,IAAI,GAAG,CAAC,CAAC,CAAA;YACf,KAAK,IAAI,IAAI,CAAA;SACd;QACD,IAAI,MAAM,GAAG,KAAK,EAAE;YAClB,CAAC,IAAI,CAAC,IAAI,GAAG,CAAC,CAAC,CAAA;YACf,MAAM,IAAI,IAAI,CAAA;SACf;QACD,OAAO,IAAI,IAAI,CAAC,CAAC,EAAE,CAAC,EAAE,KAAK,EAAE,MAAM,CAAC,CAAA;IACtC,CAAC;IAEM,kBAAG,GAAV,UAAW,IAAY,EAAE,IAAY;QAC/B,IAAA,SAA8B,EAA5B,QAAC,EAAE,QAAC,EAAE,gBAAK,EAAE,kBAAM,CAAS;QAClC,OAAO,IAAI,IAAI,CAAC,CAAC,GAAG,CAAC,IAAI,GAAG,CAAC,CAAC,EAAE,CAAC,GAAG,CAAC,IAAI,GAAG,CAAC,CAAC,EAAE,KAAK,GAAG,IAAI,EAAE,MAAM,GAAG,IAAI,CAAC,CAAA;IAC9E,CAAC;IAEM,oBAAK,GAAZ;QACE,OAAO,IAAI,IAAI,CACb,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,EAClB,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,EAClB,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,EACtB,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,CACxB,CAAA;IACH,CAAC;IACH,WAAC;AAAD,CAAC,AAxCD,IAwCC;AAxCY,oBAAI"}
\ No newline at end of file \ No newline at end of file
build/commons/NeuralNetwork.d.ts
import * as tf from '@tensorflow/tfjs-core'; import * as tf from '@tensorflow/tfjs-core';
import { ParamMapping } from './types'; import { ParamMapping } from './types';
export declare class NeuralNetwork<TNetParams> { export declare class NeuralNetwork<TNetParams> {
private _name;
protected _params: TNetParams | undefined; protected _params: TNetParams | undefined;
protected _paramMappings: ParamMapping[]; protected _paramMappings: ParamMapping[];
constructor(_name: string);
readonly params: TNetParams | undefined; readonly params: TNetParams | undefined;
readonly paramMappings: ParamMapping[]; readonly paramMappings: ParamMapping[];
getParamFromPath(paramPath: string): tf.Tensor; getParamFromPath(paramPath: string): tf.Tensor;
...@@ -21,6 +23,16 @@ export declare class NeuralNetwork<TNetParams> { ...@@ -21,6 +23,16 @@ export declare class NeuralNetwork<TNetParams> {
}[]; }[];
variable(): void; variable(): void;
freeze(): void; freeze(): void;
dispose(): void; dispose(throwOnRedispose?: boolean): void;
load(weightsOrUrl: Float32Array | string | undefined): Promise<void>;
extractWeights(weights: Float32Array): void;
private traversePropertyPath(paramPath); private traversePropertyPath(paramPath);
protected loadQuantizedParams(_: any): Promise<{
params: TNetParams;
paramMappings: ParamMapping[];
}>;
protected extractParams(_: any): {
params: TNetParams;
paramMappings: ParamMapping[];
};
} }
build/commons/NeuralNetwork.js
"use strict"; "use strict";
Object.defineProperty(exports, "__esModule", { value: true }); Object.defineProperty(exports, "__esModule", { value: true });
var tslib_1 = require("tslib");
var tf = require("@tensorflow/tfjs-core"); var tf = require("@tensorflow/tfjs-core");
var NeuralNetwork = /** @class */ (function () { var NeuralNetwork = /** @class */ (function () {
function NeuralNetwork() { function NeuralNetwork(_name) {
this._name = _name;
this._params = undefined; this._params = undefined;
this._paramMappings = []; this._paramMappings = [];
} }
...@@ -59,10 +61,44 @@ var NeuralNetwork = /** @class */ (function () { ...@@ -59,10 +61,44 @@ var NeuralNetwork = /** @class */ (function () {
_this.reassignParamFromPath(path, tf.tensor(tensor)); _this.reassignParamFromPath(path, tf.tensor(tensor));
}); });
}; };
NeuralNetwork.prototype.dispose = function () { NeuralNetwork.prototype.dispose = function (throwOnRedispose) {
this.getParamList().forEach(function (param) { return param.tensor.dispose(); }); if (throwOnRedispose === void 0) { throwOnRedispose = true; }
this.getParamList().forEach(function (param) {
if (throwOnRedispose && param.tensor.isDisposed) {
throw new Error("param tensor has already been disposed for path " + param.path);
}
param.tensor.dispose();
});
this._params = undefined; this._params = undefined;
}; };
NeuralNetwork.prototype.load = function (weightsOrUrl) {
return tslib_1.__awaiter(this, void 0, void 0, function () {
var _a, paramMappings, params;
return tslib_1.__generator(this, function (_b) {
switch (_b.label) {
case 0:
if (weightsOrUrl instanceof Float32Array) {
this.extractWeights(weightsOrUrl);
return [2 /*return*/];
}
if (weightsOrUrl && typeof weightsOrUrl !== 'string') {
throw new Error(this._name + ".load - expected model uri, or weights as Float32Array");
}
return [4 /*yield*/, this.loadQuantizedParams(weightsOrUrl)];
case 1:
_a = _b.sent(), paramMappings = _a.paramMappings, params = _a.params;
this._paramMappings = paramMappings;
this._params = params;
return [2 /*return*/];
}
});
});
};
NeuralNetwork.prototype.extractWeights = function (weights) {
var _a = this.extractParams(weights), paramMappings = _a.paramMappings, params = _a.params;
this._paramMappings = paramMappings;
this._params = params;
};
NeuralNetwork.prototype.traversePropertyPath = function (paramPath) { NeuralNetwork.prototype.traversePropertyPath = function (paramPath) {
if (!this.params) { if (!this.params) {
throw new Error("traversePropertyPath - model has no loaded params"); throw new Error("traversePropertyPath - model has no loaded params");
...@@ -79,6 +115,12 @@ var NeuralNetwork = /** @class */ (function () { ...@@ -79,6 +115,12 @@ var NeuralNetwork = /** @class */ (function () {
} }
return { obj: obj, objProp: objProp }; return { obj: obj, objProp: objProp };
}; };
NeuralNetwork.prototype.loadQuantizedParams = function (_) {
throw new Error(this._name + ".loadQuantizedParams - not implemented");
};
NeuralNetwork.prototype.extractParams = function (_) {
throw new Error(this._name + ".extractParams - not implemented");
};
return NeuralNetwork; return NeuralNetwork;
}()); }());
exports.NeuralNetwork = NeuralNetwork; exports.NeuralNetwork = NeuralNetwork;
......
build/commons/NeuralNetwork.js.map
{"version":3,"file":"NeuralNetwork.js","sourceRoot":"","sources":["../../src/commons/NeuralNetwork.ts"],"names":[],"mappings":";;AAAA,0CAA4C;AAI5C;IAAA;QAEY,YAAO,GAA2B,SAAS,CAAA;QAC3C,mBAAc,GAAmB,EAAE,CAAA;IAyE/C,CAAC;IAvEC,sBAAW,iCAAM;aAAjB;YACE,OAAO,IAAI,CAAC,OAAO,CAAA;QACrB,CAAC;;;OAAA;IAED,sBAAW,wCAAa;aAAxB;YACE,OAAO,IAAI,CAAC,cAAc,CAAA;QAC5B,CAAC;;;OAAA;IAEM,wCAAgB,GAAvB,UAAwB,SAAiB;QACjC,IAAA,yCAAuD,EAArD,YAAG,EAAE,oBAAO,CAAyC;QAC7D,OAAO,GAAG,CAAC,OAAO,CAAC,CAAA;IACrB,CAAC;IAEM,6CAAqB,GAA5B,UAA6B,SAAiB,EAAE,MAAiB;QACzD,IAAA,yCAAuD,EAArD,YAAG,EAAE,oBAAO,CAAyC;QAC7D,GAAG,CAAC,OAAO,CAAC,CAAC,OAAO,EAAE,CAAA;QACtB,GAAG,CAAC,OAAO,CAAC,GAAG,MAAM,CAAA;IACvB,CAAC;IAEM,oCAAY,GAAnB;QAAA,iBAKC;QAJC,OAAO,IAAI,CAAC,cAAc,CAAC,GAAG,CAAC,UAAC,EAAa;gBAAX,wBAAS;YAAO,OAAA,CAAC;gBACjD,IAAI,EAAE,SAAS;gBACf,MAAM,EAAE,KAAI,CAAC,gBAAgB,CAAC,SAAS,CAAC;aACzC,CAAC;QAHgD,CAGhD,CAAC,CAAA;IACL,CAAC;IAEM,0CAAkB,GAAzB;QACE,OAAO,IAAI,CAAC,YAAY,EAAE,CAAC,MAAM,CAAC,UAAA,KAAK,IAAI,OAAA,KAAK,CAAC,MAAM,YAAY,EAAE,CAAC,QAAQ,EAAnC,CAAmC,CAAC,CAAA;IACjF,CAAC;IAEM,uCAAe,GAAtB;QACE,OAAO,IAAI,CAAC,YAAY,EAAE,CAAC,MAAM,CAAC,UAAA,KAAK,IAAI,OAAA,CAAC,CAAC,KAAK,CAAC,MAAM,YAAY,EAAE,CAAC,QAAQ,CAAC,EAAtC,CAAsC,CAAC,CAAA;IACpF,CAAC;IAEM,gCAAQ,GAAf;QAAA,iBAIC;QAHC,IAAI,CAAC,eAAe,EAAE,CAAC,OAAO,CAAC,UAAC,EAAgB;gBAAd,cAAI,EAAE,kBAAM;YAC5C,KAAI,CAAC,qBAAqB,CAAC,IAAI,EAAE,EAAE,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAA;QACvD,CAAC,CAAC,CAAA;IACJ,CAAC;IAEM,8BAAM,GAAb;QAAA,iBAIC;QAHC,IAAI,CAAC,kBAAkB,EAAE,CAAC,OAAO,CAAC,UAAC,EAAgB;gBAAd,cAAI,EAAE,kBAAM;YAC/C,KAAI,CAAC,qBAAqB,CAAC,IAAI,EAAE,EAAE,CAAC,MAAM,CAAC,MAAa,CAAC,CAAC,CAAA;QAC5D,CAAC,CAAC,CAAA;IACJ,CAAC;IAEM,+BAAO,GAAd;QACE,IAAI,CAAC,YAAY,EAAE,CAAC,OAAO,CAAC,UAAA,KAAK,IAAI,OAAA,KAAK,CAAC,MAAM,CAAC,OAAO,EAAE,EAAtB,CAAsB,CAAC,CAAA;QAC5D,IAAI,CAAC,OAAO,GAAG,SAAS,CAAA;IAC1B,CAAC;IAEO,4CAAoB,GAA5B,UAA6B,SAAiB;QAC5C,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE;YAChB,MAAM,IAAI,KAAK,CAAC,mDAAmD,CAAC,CAAA;SACrE;QAED,IAAM,MAAM,GAAG,SAAS,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC,UAAC,GAAkD,EAAE,OAAO;YACrG,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,cAAc,CAAC,OAAO,CAAC,EAAE;gBACxC,MAAM,IAAI,KAAK,CAAC,0DAAwD,OAAO,mBAAc,SAAW,CAAC,CAAA;aAC1G;YAED,OAAO,EAAE,GAAG,EAAE,GAAG,CAAC,OAAO,EAAE,OAAO,SAAA,EAAE,OAAO,EAAE,GAAG,CAAC,OAAO,CAAC,OAAO,CAAC,EAAE,CAAA;QACrE,CAAC,EAAE,EAAE,OAAO,EAAE,IAAI,CAAC,MAAM,EAAE,CAAC,CAAA;QAEpB,IAAA,gBAAG,EAAE,wBAAO,CAAW;QAC/B,IAAI,CAAC,GAAG,IAAI,CAAC,OAAO,IAAI,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,YAAY,EAAE,CAAC,MAAM,CAAC,EAAE;YAC5D,MAAM,IAAI,KAAK,CAAC,gEAA8D,SAAW,CAAC,CAAA;SAC3F;QAED,OAAO,EAAE,GAAG,KAAA,EAAE,OAAO,SAAA,EAAE,CAAA;IACzB,CAAC;IACH,oBAAC;AAAD,CAAC,AA5ED,IA4EC;AA5EY,sCAAa"} {"version":3,"file":"NeuralNetwork.js","sourceRoot":"","sources":["../../src/commons/NeuralNetwork.ts"],"names":[],"mappings":";;;AAAA,0CAA4C;AAI5C;IAKE,uBAAoB,KAAa;QAAb,UAAK,GAAL,KAAK,CAAQ;QAHvB,YAAO,GAA2B,SAAS,CAAA;QAC3C,mBAAc,GAAmB,EAAE,CAAA;IAET,CAAC;IAErC,sBAAW,iCAAM;aAAjB;YACE,OAAO,IAAI,CAAC,OAAO,CAAA;QACrB,CAAC;;;OAAA;IAED,sBAAW,wCAAa;aAAxB;YACE,OAAO,IAAI,CAAC,cAAc,CAAA;QAC5B,CAAC;;;OAAA;IAEM,wCAAgB,GAAvB,UAAwB,SAAiB;QACjC,IAAA,yCAAuD,EAArD,YAAG,EAAE,oBAAO,CAAyC;QAC7D,OAAO,GAAG,CAAC,OAAO,CAAC,CAAA;IACrB,CAAC;IAEM,6CAAqB,GAA5B,UAA6B,SAAiB,EAAE,MAAiB;QACzD,IAAA,yCAAuD,EAArD,YAAG,EAAE,oBAAO,CAAyC;QAC7D,GAAG,CAAC,OAAO,CAAC,CAAC,OAAO,EAAE,CAAA;QACtB,GAAG,CAAC,OAAO,CAAC,GAAG,MAAM,CAAA;IACvB,CAAC;IAEM,oCAAY,GAAnB;QAAA,iBAKC;QAJC,OAAO,IAAI,CAAC,cAAc,CAAC,GAAG,CAAC,UAAC,EAAa;gBAAX,wBAAS;YAAO,OAAA,CAAC;gBACjD,IAAI,EAAE,SAAS;gBACf,MAAM,EAAE,KAAI,CAAC,gBAAgB,CAAC,SAAS,CAAC;aACzC,CAAC;QAHgD,CAGhD,CAAC,CAAA;IACL,CAAC;IAEM,0CAAkB,GAAzB;QACE,OAAO,IAAI,CAAC,YAAY,EAAE,CAAC,MAAM,CAAC,UAAA,KAAK,IAAI,OAAA,KAAK,CAAC,MAAM,YAAY,EAAE,CAAC,QAAQ,EAAnC,CAAmC,CAAC,CAAA;IACjF,CAAC;IAEM,uCAAe,GAAtB;QACE,OAAO,IAAI,CAAC,YAAY,EAAE,CAAC,MAAM,CAAC,UAAA,KAAK,IAAI,OAAA,CAAC,CAAC,KAAK,CAAC,MAAM,YAAY,EAAE,CAAC,QAAQ,CAAC,EAAtC,CAAsC,CAAC,CAAA;IACpF,CAAC;IAEM,gCAAQ,GAAf;QAAA,iBAIC;QAHC,IAAI,CAAC,eAAe,EAAE,CAAC,OAAO,CAAC,UAAC,EAAgB;gBAAd,cAAI,EAAE,kBAAM;YAC5C,KAAI,CAAC,qBAAqB,CAAC,IAAI,EAAE,EAAE,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAA;QACvD,CAAC,CAAC,CAAA;IACJ,CAAC;IAEM,8BAAM,GAAb;QAAA,iBAIC;QAHC,IAAI,CAAC,kBAAkB,EAAE,CAAC,OAAO,CAAC,UAAC,EAAgB;gBAAd,cAAI,EAAE,kBAAM;YAC/C,KAAI,CAAC,qBAAqB,CAAC,IAAI,EAAE,EAAE,CAAC,MAAM,CAAC,MAAa,CAAC,CAAC,CAAA;QAC5D,CAAC,CAAC,CAAA;IACJ,CAAC;IAEM,+BAAO,GAAd,UAAe,gBAAgC;QAAhC,iCAAA,EAAA,uBAAgC;QAC7C,IAAI,CAAC,YAAY,EAAE,CAAC,OAAO,CAAC,UAAA,KAAK;YAC/B,IAAI,gBAAgB,IAAI,KAAK,CAAC,MAAM,CAAC,UAAU,EAAE;gBAC/C,MAAM,IAAI,KAAK,CAAC,qDAAmD,KAAK,CAAC,IAAM,CAAC,CAAA;aACjF;YACD,KAAK,CAAC,MAAM,CAAC,OAAO,EAAE,CAAA;QACxB,CAAC,CAAC,CAAA;QACF,IAAI,CAAC,OAAO,GAAG,SAAS,CAAA;IAC1B,CAAC;IAEY,4BAAI,GAAjB,UAAkB,YAA+C;;;;;;wBAC/D,IAAI,YAAY,YAAY,YAAY,EAAE;4BACxC,IAAI,CAAC,cAAc,CAAC,YAAY,CAAC,CAAA;4BACjC,sBAAM;yBACP;wBAED,IAAI,YAAY,IAAI,OAAO,YAAY,KAAK,QAAQ,EAAE;4BACpD,MAAM,IAAI,KAAK,CAAI,IAAI,CAAC,KAAK,2DAAwD,CAAC,CAAA;yBACvF;wBAIG,qBAAM,IAAI,CAAC,mBAAmB,CAAC,YAAY,CAAC,EAAA;;wBAH1C,KAGF,SAA4C,EAF9C,aAAa,mBAAA,EACb,MAAM,YAAA;wBAGR,IAAI,CAAC,cAAc,GAAG,aAAa,CAAA;wBACnC,IAAI,CAAC,OAAO,GAAG,MAAM,CAAA;;;;;KACtB;IAEM,sCAAc,GAArB,UAAsB,OAAqB;QACnC,IAAA,gCAGyB,EAF7B,gCAAa,EACb,kBAAM,CACuB;QAE/B,IAAI,CAAC,cAAc,GAAG,aAAa,CAAA;QACnC,IAAI,CAAC,OAAO,GAAG,MAAM,CAAA;IACvB,CAAC;IAEO,4CAAoB,GAA5B,UAA6B,SAAiB;QAC5C,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE;YAChB,MAAM,IAAI,KAAK,CAAC,mDAAmD,CAAC,CAAA;SACrE;QAED,IAAM,MAAM,GAAG,SAAS,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC,UAAC,GAAkD,EAAE,OAAO;YACrG,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,cAAc,CAAC,OAAO,CAAC,EAAE;gBACxC,MAAM,IAAI,KAAK,CAAC,0DAAwD,OAAO,mBAAc,SAAW,CAAC,CAAA;aAC1G;YAED,OAAO,EAAE,GAAG,EAAE,GAAG,CAAC,OAAO,EAAE,OAAO,SAAA,EAAE,OAAO,EAAE,GAAG,CAAC,OAAO,CAAC,OAAO,CAAC,EAAE,CAAA;QACrE,CAAC,EAAE,EAAE,OAAO,EAAE,IAAI,CAAC,MAAM,EAAE,CAAC,CAAA;QAEpB,IAAA,gBAAG,EAAE,wBAAO,CAAW;QAC/B,IAAI,CAAC,GAAG,IAAI,CAAC,OAAO,IAAI,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,YAAY,EAAE,CAAC,MAAM,CAAC,EAAE;YAC5D,MAAM,IAAI,KAAK,CAAC,gEAA8D,SAAW,CAAC,CAAA;SAC3F;QAED,OAAO,EAAE,GAAG,KAAA,EAAE,OAAO,SAAA,EAAE,CAAA;IACzB,CAAC;IAES,2CAAmB,GAA7B,UAA8B,CAAM;QAClC,MAAM,IAAI,KAAK,CAAI,IAAI,CAAC,KAAK,2CAAwC,CAAC,CAAA;IACxE,CAAC;IAES,qCAAa,GAAvB,UAAwB,CAAM;QAC5B,MAAM,IAAI,KAAK,CAAI,IAAI,CAAC,KAAK,qCAAkC,CAAC,CAAA;IAClE,CAAC;IACH,oBAAC;AAAD,CAAC,AAvHD,IAuHC;AAvHY,sCAAa"}
\ No newline at end of file \ No newline at end of file
build/commons/disposeUnusedWeightTensors.d.ts 0 → 100644
import { ParamMapping } from './types';
export declare function disposeUnusedWeightTensors(weightMap: any, paramMappings: ParamMapping[]): void;
build/commons/disposeUnusedWeightTensors.js 0 → 100644
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
function disposeUnusedWeightTensors(weightMap, paramMappings) {
Object.keys(weightMap).forEach(function (path) {
if (!paramMappings.some(function (pm) { return pm.originalPath === path; })) {
weightMap[path].dispose();
}
});
}
exports.disposeUnusedWeightTensors = disposeUnusedWeightTensors;
//# sourceMappingURL=disposeUnusedWeightTensors.js.map
\ No newline at end of file
build/commons/disposeUnusedWeightTensors.js.map 0 → 100644
{"version":3,"file":"disposeUnusedWeightTensors.js","sourceRoot":"","sources":["../../src/commons/disposeUnusedWeightTensors.ts"],"names":[],"mappings":";;AAEA,oCAA2C,SAAc,EAAE,aAA6B;IACtF,MAAM,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,OAAO,CAAC,UAAA,IAAI;QACjC,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,UAAA,EAAE,IAAI,OAAA,EAAE,CAAC,YAAY,KAAK,IAAI,EAAxB,CAAwB,CAAC,EAAE;YACvD,SAAS,CAAC,IAAI,CAAC,CAAC,OAAO,EAAE,CAAA;SAC1B;IACH,CAAC,CAAC,CAAA;AACJ,CAAC;AAND,gEAMC"}
\ No newline at end of file
build/commons/extractWeightEntryFactory.d.ts 0 → 100644
import { ParamMapping } from './types';
export declare function extractWeightEntryFactory(weightMap: any, paramMappings: ParamMapping[]): <T>(originalPath: string, paramRank: number, mappedPath?: string | undefined) => T;
build/commons/extractWeightEntryFactory.js 0 → 100644
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
var isTensor_1 = require("./isTensor");
function extractWeightEntryFactory(weightMap, paramMappings) {
return function (originalPath, paramRank, mappedPath) {
var tensor = weightMap[originalPath];
if (!isTensor_1.isTensor(tensor, paramRank)) {
throw new Error("expected weightMap[" + originalPath + "] to be a Tensor" + paramRank + "D, instead have " + tensor);
}
paramMappings.push({ originalPath: originalPath, paramPath: mappedPath || originalPath });
return tensor;
};
}
exports.extractWeightEntryFactory = extractWeightEntryFactory;
//# sourceMappingURL=extractWeightEntryFactory.js.map
\ No newline at end of file
build/commons/extractWeightEntryFactory.js.map 0 → 100644
{"version":3,"file":"extractWeightEntryFactory.js","sourceRoot":"","sources":["../../src/commons/extractWeightEntryFactory.ts"],"names":[],"mappings":";;AAAA,uCAAsC;AAGtC,mCAA0C,SAAc,EAAE,aAA6B;IAErF,OAAO,UAAa,YAAoB,EAAE,SAAiB,EAAE,UAAmB;QAC9E,IAAM,MAAM,GAAG,SAAS,CAAC,YAAY,CAAC,CAAA;QAEtC,IAAI,CAAC,mBAAQ,CAAC,MAAM,EAAE,SAAS,CAAC,EAAE;YAChC,MAAM,IAAI,KAAK,CAAC,wBAAsB,YAAY,wBAAmB,SAAS,wBAAmB,MAAQ,CAAC,CAAA;SAC3G;QAED,aAAa,CAAC,IAAI,CAChB,EAAE,YAAY,cAAA,EAAE,SAAS,EAAE,UAAU,IAAI,YAAY,EAAE,CACxD,CAAA;QAED,OAAO,MAAM,CAAA;IACf,CAAC,CAAA;AAEH,CAAC;AAhBD,8DAgBC"}
\ No newline at end of file
build/faceDetectionNet/FaceDetectionNet.d.ts
import * as tf from '@tensorflow/tfjs-core'; import * as tf from '@tensorflow/tfjs-core';
import { NeuralNetwork } from '../commons/NeuralNetwork';
import { NetInput } from '../NetInput'; import { NetInput } from '../NetInput';
import { TNetInput } from '../types'; import { TNetInput } from '../types';
import { FaceDetection } from './FaceDetection'; import { FaceDetection } from './FaceDetection';
export declare class FaceDetectionNet { import { NetParams } from './types';
private _params; export declare class FaceDetectionNet extends NeuralNetwork<NetParams> {
load(weightsOrUrl?: Float32Array | string): Promise<void>; constructor();
extractWeights(weights: Float32Array): void;
forwardInput(input: NetInput): { forwardInput(input: NetInput): {
boxes: tf.Tensor<tf.Rank.R2>[]; boxes: tf.Tensor<tf.Rank.R2>[];
scores: tf.Tensor<tf.Rank.R1>[]; scores: tf.Tensor<tf.Rank.R1>[];
...@@ -15,4 +15,18 @@ export declare class FaceDetectionNet { ...@@ -15,4 +15,18 @@ export declare class FaceDetectionNet {
scores: tf.Tensor<tf.Rank.R1>[]; scores: tf.Tensor<tf.Rank.R1>[];
}>; }>;
locateFaces(input: TNetInput, minConfidence?: number, maxResults?: number): Promise<FaceDetection[]>; locateFaces(input: TNetInput, minConfidence?: number, maxResults?: number): Promise<FaceDetection[]>;
protected loadQuantizedParams(uri: string | undefined): Promise<{
params: NetParams;
paramMappings: {
originalPath?: string | undefined;
paramPath: string;
}[];
}>;
protected extractParams(weights: Float32Array): {
params: NetParams;
paramMappings: {
originalPath?: string | undefined;
paramPath: string;
}[];
};
} }
build/faceDetectionNet/FaceDetectionNet.js
...@@ -2,6 +2,7 @@ ...@@ -2,6 +2,7 @@
Object.defineProperty(exports, "__esModule", { value: true }); Object.defineProperty(exports, "__esModule", { value: true });
var tslib_1 = require("tslib"); var tslib_1 = require("tslib");
var tf = require("@tensorflow/tfjs-core"); var tf = require("@tensorflow/tfjs-core");
var NeuralNetwork_1 = require("../commons/NeuralNetwork");
var Rect_1 = require("../Rect"); var Rect_1 = require("../Rect");
var toNetInput_1 = require("../toNetInput"); var toNetInput_1 = require("../toNetInput");
var extractParams_1 = require("./extractParams"); var extractParams_1 = require("./extractParams");
...@@ -11,45 +12,22 @@ var mobileNetV1_1 = require("./mobileNetV1"); ...@@ -11,45 +12,22 @@ var mobileNetV1_1 = require("./mobileNetV1");
var nonMaxSuppression_1 = require("./nonMaxSuppression"); var nonMaxSuppression_1 = require("./nonMaxSuppression");
var outputLayer_1 = require("./outputLayer"); var outputLayer_1 = require("./outputLayer");
var predictionLayer_1 = require("./predictionLayer"); var predictionLayer_1 = require("./predictionLayer");
var FaceDetectionNet = /** @class */ (function () { var FaceDetectionNet = /** @class */ (function (_super) {
tslib_1.__extends(FaceDetectionNet, _super);
function FaceDetectionNet() { function FaceDetectionNet() {
return _super.call(this, 'FaceDetectionNet') || this;
} }
FaceDetectionNet.prototype.load = function (weightsOrUrl) {
return tslib_1.__awaiter(this, void 0, void 0, function () {
var _a;
return tslib_1.__generator(this, function (_b) {
switch (_b.label) {
case 0:
if (weightsOrUrl instanceof Float32Array) {
this.extractWeights(weightsOrUrl);
return [2 /*return*/];
}
if (weightsOrUrl && typeof weightsOrUrl !== 'string') {
throw new Error('FaceDetectionNet.load - expected model uri, or weights as Float32Array');
}
_a = this;
return [4 /*yield*/, loadQuantizedParams_1.loadQuantizedParams(weightsOrUrl)];
case 1:
_a._params = _b.sent();
return [2 /*return*/];
}
});
});
};
FaceDetectionNet.prototype.extractWeights = function (weights) {
this._params = extractParams_1.extractParams(weights);
};
FaceDetectionNet.prototype.forwardInput = function (input) { FaceDetectionNet.prototype.forwardInput = function (input) {
var _this = this; var params = this.params;
if (!this._params) { if (!params) {
throw new Error('FaceDetectionNet - load model before inference'); throw new Error('FaceDetectionNet - load model before inference');
} }
return tf.tidy(function () { return tf.tidy(function () {
var batchTensor = input.toBatchTensor(512, false); var batchTensor = input.toBatchTensor(512, false);
var x = tf.sub(tf.mul(batchTensor, tf.scalar(0.007843137718737125)), tf.scalar(1)); var x = tf.sub(tf.mul(batchTensor, tf.scalar(0.007843137718737125)), tf.scalar(1));
var features = mobileNetV1_1.mobileNetV1(x, _this._params.mobilenetv1_params); var features = mobileNetV1_1.mobileNetV1(x, params.mobilenetv1);
var _a = predictionLayer_1.predictionLayer(features.out, features.conv11, _this._params.prediction_layer_params), boxPredictions = _a.boxPredictions, classPredictions = _a.classPredictions; var _a = predictionLayer_1.predictionLayer(features.out, features.conv11, params.prediction_layer), boxPredictions = _a.boxPredictions, classPredictions = _a.classPredictions;
return outputLayer_1.outputLayer(boxPredictions, classPredictions, _this._params.output_layer_params); return outputLayer_1.outputLayer(boxPredictions, classPredictions, params.output_layer);
}); });
}; };
FaceDetectionNet.prototype.forward = function (input) { FaceDetectionNet.prototype.forward = function (input) {
...@@ -112,7 +90,13 @@ var FaceDetectionNet = /** @class */ (function () { ...@@ -112,7 +90,13 @@ var FaceDetectionNet = /** @class */ (function () {
}); });
}); });
}; };
FaceDetectionNet.prototype.loadQuantizedParams = function (uri) {
return loadQuantizedParams_1.loadQuantizedParams(uri);
};
FaceDetectionNet.prototype.extractParams = function (weights) {
return extractParams_1.extractParams(weights);
};
return FaceDetectionNet; return FaceDetectionNet;
}()); }(NeuralNetwork_1.NeuralNetwork));
exports.FaceDetectionNet = FaceDetectionNet; exports.FaceDetectionNet = FaceDetectionNet;
//# sourceMappingURL=FaceDetectionNet.js.map //# sourceMappingURL=FaceDetectionNet.js.map
\ No newline at end of file
build/faceDetectionNet/FaceDetectionNet.js.map
{"version":3,"file":"FaceDetectionNet.js","sourceRoot":"","sources":["../../src/faceDetectionNet/FaceDetectionNet.ts"],"names":[],"mappings":";;;AAAA,0CAA4C;AAG5C,gCAA+B;AAC/B,4CAA2C;AAE3C,iDAAgD;AAChD,iDAAgD;AAChD,6DAA4D;AAC5D,6CAA4C;AAC5C,yDAAwD;AACxD,6CAA4C;AAC5C,qDAAoD;AAGpD;IAAA;IAgHA,CAAC;IA5Gc,+BAAI,GAAjB,UAAkB,YAAoC;;;;;;wBACpD,IAAI,YAAY,YAAY,YAAY,EAAE;4BACxC,IAAI,CAAC,cAAc,CAAC,YAAY,CAAC,CAAA;4BACjC,sBAAM;yBACP;wBAED,IAAI,YAAY,IAAI,OAAO,YAAY,KAAK,QAAQ,EAAE;4BACpD,MAAM,IAAI,KAAK,CAAC,wEAAwE,CAAC,CAAA;yBAC1F;wBACD,KAAA,IAAI,CAAA;wBAAW,qBAAM,yCAAmB,CAAC,YAAY,CAAC,EAAA;;wBAAtD,GAAK,OAAO,GAAG,SAAuC,CAAA;;;;;KACvD;IAEM,yCAAc,GAArB,UAAsB,OAAqB;QACzC,IAAI,CAAC,OAAO,GAAG,6BAAa,CAAC,OAAO,CAAC,CAAA;IACvC,CAAC;IAEM,uCAAY,GAAnB,UAAoB,KAAe;QAAnC,iBAkBC;QAjBC,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE;YACjB,MAAM,IAAI,KAAK,CAAC,gDAAgD,CAAC,CAAA;SAClE;QAED,OAAO,EAAE,CAAC,IAAI,CAAC;YACb,IAAM,WAAW,GAAG,KAAK,CAAC,aAAa,CAAC,GAAG,EAAE,KAAK,CAAC,CAAA;YAEnD,IAAM,CAAC,GAAG,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,WAAW,EAAE,EAAE,CAAC,MAAM,CAAC,oBAAoB,CAAC,CAAC,EAAE,EAAE,CAAC,MAAM,CAAC,CAAC,CAAC,CAAgB,CAAA;YACnG,IAAM,QAAQ,GAAG,yBAAW,CAAC,CAAC,EAAE,KAAI,CAAC,OAAO,CAAC,kBAAkB,CAAC,CAAA;YAE1D,IAAA,4GAGkF,EAFtF,kCAAc,EACd,sCAAgB,CACsE;YAExF,OAAO,yBAAW,CAAC,cAAc,EAAE,gBAAgB,EAAE,KAAI,CAAC,OAAO,CAAC,mBAAmB,CAAC,CAAA;QACxF,CAAC,CAAC,CAAA;IACJ,CAAC;IAEY,kCAAO,GAApB,UAAqB,KAAgB;;;;;;wBAC5B,KAAA,IAAI,CAAC,YAAY,CAAA;wBAAC,qBAAM,uBAAU,CAAC,KAAK,EAAE,IAAI,CAAC,EAAA;4BAAtD,sBAAO,SAAA,IAAI,GAAc,SAA6B,EAAC,EAAA;;;;KACxD;IAEY,sCAAW,GAAxB,UACE,KAAgB,EAChB,aAA2B,EAC3B,UAAwB;QADxB,8BAAA,EAAA,mBAA2B;QAC3B,2BAAA,EAAA,gBAAwB;;;;;4BAGP,qBAAM,uBAAU,CAAC,KAAK,EAAE,IAAI,CAAC,EAAA;;wBAAxC,QAAQ,GAAG,SAA6B;wBAExC,KAGF,IAAI,CAAC,YAAY,CAAC,QAAQ,CAAC,EAFtB,MAAM,WAAA,EACL,OAAO,YAAA,CACc;wBAIzB,KAAK,GAAG,MAAM,CAAC,CAAC,CAAC,CAAA;wBACjB,MAAM,GAAG,OAAO,CAAC,CAAC,CAAC,CAAA;wBACzB,KAAS,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;4BACtC,MAAM,CAAC,CAAC,CAAC,CAAC,OAAO,EAAE,CAAA;4BACnB,OAAO,CAAC,CAAC,CAAC,CAAC,OAAO,EAAE,CAAA;yBACrB;wBAGkB,KAAA,CAAA,KAAA,KAAK,CAAA,CAAC,IAAI,CAAA;wBAAC,qBAAM,MAAM,CAAC,IAAI,EAAE,EAAA;;wBAA3C,UAAU,GAAG,cAAW,SAAmB,EAAC;wBAE5C,YAAY,GAAG,GAAG,CAAA;wBAClB,OAAO,GAAG,qCAAiB,CAC/B,KAAK,EACL,UAAU,EACV,UAAU,EACV,YAAY,EACZ,aAAa,CACd,CAAA;wBAGK,oBAAoB,GAAG,CAAC,QAAQ,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,QAAQ,CAAC,cAAc,CAAC,CAAC,CAAC,CAAC,GAAG,QAAQ,CAAC,cAAc,CAAC,CAAC,CAAC,CAAA;wBAC5G,mBAAmB,GAAG,CAAC,QAAQ,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,QAAQ,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,GAAG,QAAQ,CAAC,aAAa,CAAC,CAAC,CAAC,CAAA;wBAEzG,OAAO,GAAG,OAAO;6BACpB,GAAG,CAAC,UAAA,GAAG;4BACA,IAAA;;;wFAGkC,EAHjC,WAAG,EAAE,cAAM,CAGsB;4BAClC,IAAA;;;uFAGiC,EAHhC,YAAI,EAAE,aAAK,CAGqB;4BACvC,OAAO,IAAI,6BAAa,CACtB,UAAU,CAAC,GAAG,CAAC,EACf,IAAI,WAAI,CACN,IAAI,EACJ,GAAG,EACH,KAAK,GAAG,IAAI,EACZ,MAAM,GAAG,GAAG,CACb,EACD;gCACE,MAAM,EAAE,QAAQ,CAAC,cAAc,CAAC,CAAC,CAAC;gCAClC,KAAK,EAAE,QAAQ,CAAC,aAAa,CAAC,CAAC,CAAC;6BACjC,CACF,CAAA;wBACH,CAAC,CAAC,CAAA;wBAEJ,KAAK,CAAC,OAAO,EAAE,CAAA;wBACf,MAAM,CAAC,OAAO,EAAE,CAAA;wBAEhB,sBAAO,OAAO,EAAA;;;;KACf;IACH,uBAAC;AAAD,CAAC,AAhHD,IAgHC;AAhHY,4CAAgB"} {"version":3,"file":"FaceDetectionNet.js","sourceRoot":"","sources":["../../src/faceDetectionNet/FaceDetectionNet.ts"],"names":[],"mappings":";;;AAAA,0CAA4C;AAE5C,0DAAyD;AAEzD,gCAA+B;AAC/B,4CAA2C;AAE3C,iDAAgD;AAChD,iDAAgD;AAChD,6DAA4D;AAC5D,6CAA4C;AAC5C,yDAAwD;AACxD,6CAA4C;AAC5C,qDAAoD;AAGpD;IAAsC,4CAAwB;IAE5D;eACE,kBAAM,kBAAkB,CAAC;IAC3B,CAAC;IAEM,uCAAY,GAAnB,UAAoB,KAAe;QAEzB,IAAA,oBAAM,CAAS;QAEvB,IAAI,CAAC,MAAM,EAAE;YACX,MAAM,IAAI,KAAK,CAAC,gDAAgD,CAAC,CAAA;SAClE;QAED,OAAO,EAAE,CAAC,IAAI,CAAC;YACb,IAAM,WAAW,GAAG,KAAK,CAAC,aAAa,CAAC,GAAG,EAAE,KAAK,CAAC,CAAA;YAEnD,IAAM,CAAC,GAAG,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,WAAW,EAAE,EAAE,CAAC,MAAM,CAAC,oBAAoB,CAAC,CAAC,EAAE,EAAE,CAAC,MAAM,CAAC,CAAC,CAAC,CAAgB,CAAA;YACnG,IAAM,QAAQ,GAAG,yBAAW,CAAC,CAAC,EAAE,MAAM,CAAC,WAAW,CAAC,CAAA;YAE7C,IAAA,8FAGqE,EAFzE,kCAAc,EACd,sCAAgB,CACyD;YAE3E,OAAO,yBAAW,CAAC,cAAc,EAAE,gBAAgB,EAAE,MAAM,CAAC,YAAY,CAAC,CAAA;QAC3E,CAAC,CAAC,CAAA;IACJ,CAAC;IAEY,kCAAO,GAApB,UAAqB,KAAgB;;;;;;wBAC5B,KAAA,IAAI,CAAC,YAAY,CAAA;wBAAC,qBAAM,uBAAU,CAAC,KAAK,EAAE,IAAI,CAAC,EAAA;4BAAtD,sBAAO,SAAA,IAAI,GAAc,SAA6B,EAAC,EAAA;;;;KACxD;IAEY,sCAAW,GAAxB,UACE,KAAgB,EAChB,aAA2B,EAC3B,UAAwB;QADxB,8BAAA,EAAA,mBAA2B;QAC3B,2BAAA,EAAA,gBAAwB;;;;;4BAGP,qBAAM,uBAAU,CAAC,KAAK,EAAE,IAAI,CAAC,EAAA;;wBAAxC,QAAQ,GAAG,SAA6B;wBAExC,KAGF,IAAI,CAAC,YAAY,CAAC,QAAQ,CAAC,EAFtB,MAAM,WAAA,EACL,OAAO,YAAA,CACc;wBAIzB,KAAK,GAAG,MAAM,CAAC,CAAC,CAAC,CAAA;wBACjB,MAAM,GAAG,OAAO,CAAC,CAAC,CAAC,CAAA;wBACzB,KAAS,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;4BACtC,MAAM,CAAC,CAAC,CAAC,CAAC,OAAO,EAAE,CAAA;4BACnB,OAAO,CAAC,CAAC,CAAC,CAAC,OAAO,EAAE,CAAA;yBACrB;wBAGkB,KAAA,CAAA,KAAA,KAAK,CAAA,CAAC,IAAI,CAAA;wBAAC,qBAAM,MAAM,CAAC,IAAI,EAAE,EAAA;;wBAA3C,UAAU,GAAG,cAAW,SAAmB,EAAC;wBAE5C,YAAY,GAAG,GAAG,CAAA;wBAClB,OAAO,GAAG,qCAAiB,CAC/B,KAAK,EACL,UAAU,EACV,UAAU,EACV,YAAY,EACZ,aAAa,CACd,CAAA;wBAEK,oBAAoB,GAAG,CAAC,QAAQ,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,QAAQ,CAAC,cAAc,CAAC,CAAC,CAAC,CAAC,GAAG,QAAQ,CAAC,cAAc,CAAC,CAAC,CAAC,CAAA;wBAC5G,mBAAmB,GAAG,CAAC,QAAQ,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,QAAQ,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,GAAG,QAAQ,CAAC,aAAa,CAAC,CAAC,CAAC,CAAA;wBAEzG,OAAO,GAAG,OAAO;6BACpB,GAAG,CAAC,UAAA,GAAG;4BACA,IAAA;;;wFAGkC,EAHjC,WAAG,EAAE,cAAM,CAGsB;4BAClC,IAAA;;;uFAGiC,EAHhC,YAAI,EAAE,aAAK,CAGqB;4BACvC,OAAO,IAAI,6BAAa,CACtB,UAAU,CAAC,GAAG,CAAC,EACf,IAAI,WAAI,CACN,IAAI,EACJ,GAAG,EACH,KAAK,GAAG,IAAI,EACZ,MAAM,GAAG,GAAG,CACb,EACD;gCACE,MAAM,EAAE,QAAQ,CAAC,cAAc,CAAC,CAAC,CAAC;gCAClC,KAAK,EAAE,QAAQ,CAAC,aAAa,CAAC,CAAC,CAAC;6BACjC,CACF,CAAA;wBACH,CAAC,CAAC,CAAA;wBAEJ,KAAK,CAAC,OAAO,EAAE,CAAA;wBACf,MAAM,CAAC,OAAO,EAAE,CAAA;wBAEhB,sBAAO,OAAO,EAAA;;;;KACf;IAES,8CAAmB,GAA7B,UAA8B,GAAuB;QACnD,OAAO,yCAAmB,CAAC,GAAG,CAAC,CAAA;IACjC,CAAC;IAES,wCAAa,GAAvB,UAAwB,OAAqB;QAC3C,OAAO,6BAAa,CAAC,OAAO,CAAC,CAAA;IAC/B,CAAC;IACH,uBAAC;AAAD,CAAC,AA5GD,CAAsC,6BAAa,GA4GlD;AA5GY,4CAAgB"}
\ No newline at end of file \ No newline at end of file
build/faceDetectionNet/boxPredictionLayer.js
...@@ -5,8 +5,8 @@ var convLayer_1 = require("../commons/convLayer"); ...@@ -5,8 +5,8 @@ var convLayer_1 = require("../commons/convLayer");
function boxPredictionLayer(x, params) { function boxPredictionLayer(x, params) {
return tf.tidy(function () { return tf.tidy(function () {
var batchSize = x.shape[0]; var batchSize = x.shape[0];
var boxPredictionEncoding = tf.reshape(convLayer_1.convLayer(x, params.box_encoding_predictor_params), [batchSize, -1, 1, 4]); var boxPredictionEncoding = tf.reshape(convLayer_1.convLayer(x, params.box_encoding_predictor), [batchSize, -1, 1, 4]);
var classPrediction = tf.reshape(convLayer_1.convLayer(x, params.class_predictor_params), [batchSize, -1, 3]); var classPrediction = tf.reshape(convLayer_1.convLayer(x, params.class_predictor), [batchSize, -1, 3]);
return { return {
boxPredictionEncoding: boxPredictionEncoding, boxPredictionEncoding: boxPredictionEncoding,
classPrediction: classPrediction classPrediction: classPrediction
......
build/faceDetectionNet/boxPredictionLayer.js.map
{"version":3,"file":"boxPredictionLayer.js","sourceRoot":"","sources":["../../src/faceDetectionNet/boxPredictionLayer.ts"],"names":[],"mappings":";;AAAA,0CAA4C;AAE5C,kDAAiD;AAIjD,4BACE,CAAc,EACd,MAA2B;IAE3B,OAAO,EAAE,CAAC,IAAI,CAAC;QAEb,IAAM,SAAS,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAA;QAE5B,IAAM,qBAAqB,GAAG,EAAE,CAAC,OAAO,CACtC,qBAAS,CAAC,CAAC,EAAE,MAAM,CAAC,6BAA6B,CAAC,EAClD,CAAC,SAAS,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,CACtB,CAAA;QACD,IAAM,eAAe,GAAG,EAAE,CAAC,OAAO,CAChC,qBAAS,CAAC,CAAC,EAAE,MAAM,CAAC,sBAAsB,CAAC,EAC3C,CAAC,SAAS,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CACnB,CAAA;QAED,OAAO;YACL,qBAAqB,uBAAA;YACrB,eAAe,iBAAA;SAChB,CAAA;IACH,CAAC,CAAC,CAAA;AACJ,CAAC;AAtBD,gDAsBC"} {"version":3,"file":"boxPredictionLayer.js","sourceRoot":"","sources":["../../src/faceDetectionNet/boxPredictionLayer.ts"],"names":[],"mappings":";;AAAA,0CAA4C;AAE5C,kDAAiD;AAIjD,4BACE,CAAc,EACd,MAA2B;IAE3B,OAAO,EAAE,CAAC,IAAI,CAAC;QAEb,IAAM,SAAS,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAA;QAE5B,IAAM,qBAAqB,GAAG,EAAE,CAAC,OAAO,CACtC,qBAAS,CAAC,CAAC,EAAE,MAAM,CAAC,sBAAsB,CAAC,EAC3C,CAAC,SAAS,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,CACtB,CAAA;QACD,IAAM,eAAe,GAAG,EAAE,CAAC,OAAO,CAChC,qBAAS,CAAC,CAAC,EAAE,MAAM,CAAC,eAAe,CAAC,EACpC,CAAC,SAAS,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CACnB,CAAA;QAED,OAAO;YACL,qBAAqB,uBAAA;YACrB,eAAe,iBAAA;SAChB,CAAA;IACH,CAAC,CAAC,CAAA;AACJ,CAAC;AAtBD,gDAsBC"}
\ No newline at end of file \ No newline at end of file
build/faceDetectionNet/extractParams.d.ts
import { ParamMapping } from '../commons/types';
import { NetParams } from './types'; import { NetParams } from './types';
export declare function extractParams(weights: Float32Array): NetParams; export declare function extractParams(weights: Float32Array): {
params: NetParams;
paramMappings: ParamMapping[];
};
build/faceDetectionNet/extractParams.js
...@@ -2,13 +2,14 @@ ...@@ -2,13 +2,14 @@
Object.defineProperty(exports, "__esModule", { value: true }); Object.defineProperty(exports, "__esModule", { value: true });
var tf = require("@tensorflow/tfjs-core"); var tf = require("@tensorflow/tfjs-core");
var extractWeightsFactory_1 = require("../commons/extractWeightsFactory"); var extractWeightsFactory_1 = require("../commons/extractWeightsFactory");
function extractorsFactory(extractWeights) { function extractorsFactory(extractWeights, paramMappings) {
function extractDepthwiseConvParams(numChannels) { function extractDepthwiseConvParams(numChannels, mappedPrefix) {
var filters = tf.tensor4d(extractWeights(3 * 3 * numChannels), [3, 3, numChannels, 1]); var filters = tf.tensor4d(extractWeights(3 * 3 * numChannels), [3, 3, numChannels, 1]);
var batch_norm_scale = tf.tensor1d(extractWeights(numChannels)); var batch_norm_scale = tf.tensor1d(extractWeights(numChannels));
var batch_norm_offset = tf.tensor1d(extractWeights(numChannels)); var batch_norm_offset = tf.tensor1d(extractWeights(numChannels));
var batch_norm_mean = tf.tensor1d(extractWeights(numChannels)); var batch_norm_mean = tf.tensor1d(extractWeights(numChannels));
var batch_norm_variance = tf.tensor1d(extractWeights(numChannels)); var batch_norm_variance = tf.tensor1d(extractWeights(numChannels));
paramMappings.push({ paramPath: mappedPrefix + "/filters" }, { paramPath: mappedPrefix + "/batch_norm_scale" }, { paramPath: mappedPrefix + "/batch_norm_offset" }, { paramPath: mappedPrefix + "/batch_norm_mean" }, { paramPath: mappedPrefix + "/batch_norm_variance" });
return { return {
filters: filters, filters: filters,
batch_norm_scale: batch_norm_scale, batch_norm_scale: batch_norm_scale,
...@@ -17,115 +18,116 @@ function extractorsFactory(extractWeights) { ...@@ -17,115 +18,116 @@ function extractorsFactory(extractWeights) {
batch_norm_variance: batch_norm_variance batch_norm_variance: batch_norm_variance
}; };
} }
function extractConvParams(channelsIn, channelsOut, filterSize) { function extractConvParams(channelsIn, channelsOut, filterSize, mappedPrefix, isPointwiseConv) {
var filters = tf.tensor4d(extractWeights(channelsIn * channelsOut * filterSize * filterSize), [filterSize, filterSize, channelsIn, channelsOut]); var filters = tf.tensor4d(extractWeights(channelsIn * channelsOut * filterSize * filterSize), [filterSize, filterSize, channelsIn, channelsOut]);
var bias = tf.tensor1d(extractWeights(channelsOut)); var bias = tf.tensor1d(extractWeights(channelsOut));
return { paramMappings.push({ paramPath: mappedPrefix + "/filters" }, { paramPath: mappedPrefix + "/" + (isPointwiseConv ? 'batch_norm_offset' : 'bias') });
filters: filters, return { filters: filters, bias: bias };
bias: bias
};
} }
function extractPointwiseConvParams(channelsIn, channelsOut, filterSize) { function extractPointwiseConvParams(channelsIn, channelsOut, filterSize, mappedPrefix) {
var _a = extractConvParams(channelsIn, channelsOut, filterSize), filters = _a.filters, bias = _a.bias; var _a = extractConvParams(channelsIn, channelsOut, filterSize, mappedPrefix, true), filters = _a.filters, bias = _a.bias;
return { return {
filters: filters, filters: filters,
batch_norm_offset: bias batch_norm_offset: bias
}; };
} }
function extractConvPairParams(channelsIn, channelsOut) { function extractConvPairParams(channelsIn, channelsOut, mappedPrefix) {
var depthwise_conv_params = extractDepthwiseConvParams(channelsIn); var depthwise_conv = extractDepthwiseConvParams(channelsIn, mappedPrefix + "/depthwise_conv");
var pointwise_conv_params = extractPointwiseConvParams(channelsIn, channelsOut, 1); var pointwise_conv = extractPointwiseConvParams(channelsIn, channelsOut, 1, mappedPrefix + "/pointwise_conv");
return { return { depthwise_conv: depthwise_conv, pointwise_conv: pointwise_conv };
depthwise_conv_params: depthwise_conv_params,
pointwise_conv_params: pointwise_conv_params
};
} }
function extractMobilenetV1Params() { function extractMobilenetV1Params() {
var conv_0_params = extractPointwiseConvParams(3, 32, 3); var conv_0 = extractPointwiseConvParams(3, 32, 3, 'mobilenetv1/conv_0');
var channelNumPairs = [ var conv_1 = extractConvPairParams(32, 64, 'mobilenetv1/conv_1');
[32, 64], var conv_2 = extractConvPairParams(64, 128, 'mobilenetv1/conv_2');
[64, 128], var conv_3 = extractConvPairParams(128, 128, 'mobilenetv1/conv_3');
[128, 128], var conv_4 = extractConvPairParams(128, 256, 'mobilenetv1/conv_4');
[128, 256], var conv_5 = extractConvPairParams(256, 256, 'mobilenetv1/conv_5');
[256, 256], var conv_6 = extractConvPairParams(256, 512, 'mobilenetv1/conv_6');
[256, 512], var conv_7 = extractConvPairParams(512, 512, 'mobilenetv1/conv_7');
[512, 512], var conv_8 = extractConvPairParams(512, 512, 'mobilenetv1/conv_8');
[512, 512], var conv_9 = extractConvPairParams(512, 512, 'mobilenetv1/conv_9');
[512, 512], var conv_10 = extractConvPairParams(512, 512, 'mobilenetv1/conv_10');
[512, 512], var conv_11 = extractConvPairParams(512, 512, 'mobilenetv1/conv_11');
[512, 512], var conv_12 = extractConvPairParams(512, 1024, 'mobilenetv1/conv_12');
[512, 1024], var conv_13 = extractConvPairParams(1024, 1024, 'mobilenetv1/conv_13');
[1024, 1024]
];
var conv_pair_params = channelNumPairs.map(function (_a) {
var channelsIn = _a[0], channelsOut = _a[1];
return extractConvPairParams(channelsIn, channelsOut);
});
return { return {
conv_0_params: conv_0_params, conv_0: conv_0,
conv_pair_params: conv_pair_params conv_1: conv_1,
conv_2: conv_2,
conv_3: conv_3,
conv_4: conv_4,
conv_5: conv_5,
conv_6: conv_6,
conv_7: conv_7,
conv_8: conv_8,
conv_9: conv_9,
conv_10: conv_10,
conv_11: conv_11,
conv_12: conv_12,
conv_13: conv_13
}; };
} }
function extractPredictionLayerParams() { function extractPredictionLayerParams() {
var conv_0_params = extractPointwiseConvParams(1024, 256, 1); var conv_0 = extractPointwiseConvParams(1024, 256, 1, 'prediction_layer/conv_0');
var conv_1_params = extractPointwiseConvParams(256, 512, 3); var conv_1 = extractPointwiseConvParams(256, 512, 3, 'prediction_layer/conv_1');
var conv_2_params = extractPointwiseConvParams(512, 128, 1); var conv_2 = extractPointwiseConvParams(512, 128, 1, 'prediction_layer/conv_2');
var conv_3_params = extractPointwiseConvParams(128, 256, 3); var conv_3 = extractPointwiseConvParams(128, 256, 3, 'prediction_layer/conv_3');
var conv_4_params = extractPointwiseConvParams(256, 128, 1); var conv_4 = extractPointwiseConvParams(256, 128, 1, 'prediction_layer/conv_4');
var conv_5_params = extractPointwiseConvParams(128, 256, 3); var conv_5 = extractPointwiseConvParams(128, 256, 3, 'prediction_layer/conv_5');
var conv_6_params = extractPointwiseConvParams(256, 64, 1); var conv_6 = extractPointwiseConvParams(256, 64, 1, 'prediction_layer/conv_6');
var conv_7_params = extractPointwiseConvParams(64, 128, 3); var conv_7 = extractPointwiseConvParams(64, 128, 3, 'prediction_layer/conv_7');
var box_encoding_0_predictor_params = extractConvParams(512, 12, 1); var box_encoding_0_predictor = extractConvParams(512, 12, 1, 'prediction_layer/box_predictor_0/box_encoding_predictor');
var class_predictor_0_params = extractConvParams(512, 9, 1); var class_predictor_0 = extractConvParams(512, 9, 1, 'prediction_layer/box_predictor_0/class_predictor');
var box_encoding_1_predictor_params = extractConvParams(1024, 24, 1); var box_encoding_1_predictor = extractConvParams(1024, 24, 1, 'prediction_layer/box_predictor_1/box_encoding_predictor');
var class_predictor_1_params = extractConvParams(1024, 18, 1); var class_predictor_1 = extractConvParams(1024, 18, 1, 'prediction_layer/box_predictor_1/class_predictor');
var box_encoding_2_predictor_params = extractConvParams(512, 24, 1); var box_encoding_2_predictor = extractConvParams(512, 24, 1, 'prediction_layer/box_predictor_2/box_encoding_predictor');
var class_predictor_2_params = extractConvParams(512, 18, 1); var class_predictor_2 = extractConvParams(512, 18, 1, 'prediction_layer/box_predictor_2/class_predictor');
var box_encoding_3_predictor_params = extractConvParams(256, 24, 1); var box_encoding_3_predictor = extractConvParams(256, 24, 1, 'prediction_layer/box_predictor_3/box_encoding_predictor');
var class_predictor_3_params = extractConvParams(256, 18, 1); var class_predictor_3 = extractConvParams(256, 18, 1, 'prediction_layer/box_predictor_3/class_predictor');
var box_encoding_4_predictor_params = extractConvParams(256, 24, 1); var box_encoding_4_predictor = extractConvParams(256, 24, 1, 'prediction_layer/box_predictor_4/box_encoding_predictor');
var class_predictor_4_params = extractConvParams(256, 18, 1); var class_predictor_4 = extractConvParams(256, 18, 1, 'prediction_layer/box_predictor_4/class_predictor');
var box_encoding_5_predictor_params = extractConvParams(128, 24, 1); var box_encoding_5_predictor = extractConvParams(128, 24, 1, 'prediction_layer/box_predictor_5/box_encoding_predictor');
var class_predictor_5_params = extractConvParams(128, 18, 1); var class_predictor_5 = extractConvParams(128, 18, 1, 'prediction_layer/box_predictor_5/class_predictor');
var box_predictor_0_params = { var box_predictor_0 = {
box_encoding_predictor_params: box_encoding_0_predictor_params, box_encoding_predictor: box_encoding_0_predictor,
class_predictor_params: class_predictor_0_params class_predictor: class_predictor_0
}; };
var box_predictor_1_params = { var box_predictor_1 = {
box_encoding_predictor_params: box_encoding_1_predictor_params, box_encoding_predictor: box_encoding_1_predictor,
class_predictor_params: class_predictor_1_params class_predictor: class_predictor_1
}; };
var box_predictor_2_params = { var box_predictor_2 = {
box_encoding_predictor_params: box_encoding_2_predictor_params, box_encoding_predictor: box_encoding_2_predictor,
class_predictor_params: class_predictor_2_params class_predictor: class_predictor_2
}; };
var box_predictor_3_params = { var box_predictor_3 = {
box_encoding_predictor_params: box_encoding_3_predictor_params, box_encoding_predictor: box_encoding_3_predictor,
class_predictor_params: class_predictor_3_params class_predictor: class_predictor_3
}; };
var box_predictor_4_params = { var box_predictor_4 = {
box_encoding_predictor_params: box_encoding_4_predictor_params, box_encoding_predictor: box_encoding_4_predictor,
class_predictor_params: class_predictor_4_params class_predictor: class_predictor_4
}; };
var box_predictor_5_params = { var box_predictor_5 = {
box_encoding_predictor_params: box_encoding_5_predictor_params, box_encoding_predictor: box_encoding_5_predictor,
class_predictor_params: class_predictor_5_params class_predictor: class_predictor_5
}; };
return { return {
conv_0_params: conv_0_params, conv_0: conv_0,
conv_1_params: conv_1_params, conv_1: conv_1,
conv_2_params: conv_2_params, conv_2: conv_2,
conv_3_params: conv_3_params, conv_3: conv_3,
conv_4_params: conv_4_params, conv_4: conv_4,
conv_5_params: conv_5_params, conv_5: conv_5,
conv_6_params: conv_6_params, conv_6: conv_6,
conv_7_params: conv_7_params, conv_7: conv_7,
box_predictor_0_params: box_predictor_0_params, box_predictor_0: box_predictor_0,
box_predictor_1_params: box_predictor_1_params, box_predictor_1: box_predictor_1,
box_predictor_2_params: box_predictor_2_params, box_predictor_2: box_predictor_2,
box_predictor_3_params: box_predictor_3_params, box_predictor_3: box_predictor_3,
box_predictor_4_params: box_predictor_4_params, box_predictor_4: box_predictor_4,
box_predictor_5_params: box_predictor_5_params box_predictor_5: box_predictor_5
}; };
} }
return { return {
...@@ -134,21 +136,26 @@ function extractorsFactory(extractWeights) { ...@@ -134,21 +136,26 @@ function extractorsFactory(extractWeights) {
}; };
} }
function extractParams(weights) { function extractParams(weights) {
var paramMappings = [];
var _a = extractWeightsFactory_1.extractWeightsFactory(weights), extractWeights = _a.extractWeights, getRemainingWeights = _a.getRemainingWeights; var _a = extractWeightsFactory_1.extractWeightsFactory(weights), extractWeights = _a.extractWeights, getRemainingWeights = _a.getRemainingWeights;
var _b = extractorsFactory(extractWeights), extractMobilenetV1Params = _b.extractMobilenetV1Params, extractPredictionLayerParams = _b.extractPredictionLayerParams; var _b = extractorsFactory(extractWeights, paramMappings), extractMobilenetV1Params = _b.extractMobilenetV1Params, extractPredictionLayerParams = _b.extractPredictionLayerParams;
var mobilenetv1_params = extractMobilenetV1Params(); var mobilenetv1 = extractMobilenetV1Params();
var prediction_layer_params = extractPredictionLayerParams(); var prediction_layer = extractPredictionLayerParams();
var extra_dim = tf.tensor3d(extractWeights(5118 * 4), [1, 5118, 4]); var extra_dim = tf.tensor3d(extractWeights(5118 * 4), [1, 5118, 4]);
var output_layer_params = { var output_layer = {
extra_dim: extra_dim extra_dim: extra_dim
}; };
paramMappings.push({ paramPath: 'output_layer/extra_dim' });
if (getRemainingWeights().length !== 0) { if (getRemainingWeights().length !== 0) {
throw new Error("weights remaing after extract: " + getRemainingWeights().length); throw new Error("weights remaing after extract: " + getRemainingWeights().length);
} }
return { return {
mobilenetv1_params: mobilenetv1_params, params: {
prediction_layer_params: prediction_layer_params, mobilenetv1: mobilenetv1,
output_layer_params: output_layer_params prediction_layer: prediction_layer,
output_layer: output_layer
},
paramMappings: paramMappings
}; };
} }
exports.extractParams = extractParams; exports.extractParams = extractParams;
......
build/faceDetectionNet/extractParams.js.map
{"version":3,"file":"extractParams.js","sourceRoot":"","sources":["../../src/faceDetectionNet/extractParams.ts"],"names":[],"mappings":";;AAAA,0CAA4C;AAE5C,0EAAyE;AAIzE,2BAA2B,cAAoD;IAE7E,oCAAoC,WAAmB;QACrD,IAAM,OAAO,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,CAAC,GAAG,CAAC,GAAG,WAAW,CAAC,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,WAAW,EAAE,CAAC,CAAC,CAAC,CAAA;QACxF,IAAM,gBAAgB,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,WAAW,CAAC,CAAC,CAAA;QACjE,IAAM,iBAAiB,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,WAAW,CAAC,CAAC,CAAA;QAClE,IAAM,eAAe,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,WAAW,CAAC,CAAC,CAAA;QAChE,IAAM,mBAAmB,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,WAAW,CAAC,CAAC,CAAA;QAEpE,OAAO;YACL,OAAO,SAAA;YACP,gBAAgB,kBAAA;YAChB,iBAAiB,mBAAA;YACjB,eAAe,iBAAA;YACf,mBAAmB,qBAAA;SACpB,CAAA;IACH,CAAC;IAED,2BACE,UAAkB,EAClB,WAAmB,EACnB,UAAkB;QAElB,IAAM,OAAO,GAAG,EAAE,CAAC,QAAQ,CACzB,cAAc,CAAC,UAAU,GAAG,WAAW,GAAG,UAAU,GAAG,UAAU,CAAC,EAClE,CAAC,UAAU,EAAE,UAAU,EAAE,UAAU,EAAE,WAAW,CAAC,CAClD,CAAA;QACD,IAAM,IAAI,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,WAAW,CAAC,CAAC,CAAA;QAErD,OAAO;YACL,OAAO,SAAA;YACP,IAAI,MAAA;SACL,CAAA;IACH,CAAC;IAED,oCACE,UAAkB,EAClB,WAAmB,EACnB,UAAkB;QAEZ,IAAA,2DAGoD,EAFxD,oBAAO,EACP,cAAI,CACoD;QAE1D,OAAO;YACL,OAAO,SAAA;YACP,iBAAiB,EAAE,IAAI;SACxB,CAAA;IACH,CAAC;IAED,+BACE,UAAkB,EAClB,WAAmB;QAEnB,IAAM,qBAAqB,GAAG,0BAA0B,CAAC,UAAU,CAAC,CAAA;QACpE,IAAM,qBAAqB,GAAG,0BAA0B,CAAC,UAAU,EAAE,WAAW,EAAE,CAAC,CAAC,CAAA;QAEpF,OAAO;YACL,qBAAqB,uBAAA;YACrB,qBAAqB,uBAAA;SACtB,CAAA;IACH,CAAC;IAED;QAEE,IAAM,aAAa,GAAG,0BAA0B,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;QAE1D,IAAM,eAAe,GAAG;YACtB,CAAC,EAAE,EAAE,EAAE,CAAC;YACR,CAAC,EAAE,EAAE,GAAG,CAAC;YACT,CAAC,GAAG,EAAE,GAAG,CAAC;YACV,CAAC,GAAG,EAAE,GAAG,CAAC;YACV,CAAC,GAAG,EAAE,GAAG,CAAC;YACV,CAAC,GAAG,EAAE,GAAG,CAAC;YACV,CAAC,GAAG,EAAE,GAAG,CAAC;YACV,CAAC,GAAG,EAAE,GAAG,CAAC;YACV,CAAC,GAAG,EAAE,GAAG,CAAC;YACV,CAAC,GAAG,EAAE,GAAG,CAAC;YACV,CAAC,GAAG,EAAE,GAAG,CAAC;YACV,CAAC,GAAG,EAAE,IAAI,CAAC;YACX,CAAC,IAAI,EAAE,IAAI,CAAC;SACb,CAAA;QAED,IAAM,gBAAgB,GAAG,eAAe,CAAC,GAAG,CAC1C,UAAC,EAAyB;gBAAxB,kBAAU,EAAE,mBAAW;YAAM,OAAA,qBAAqB,CAAC,UAAU,EAAE,WAAW,CAAC;QAA9C,CAA8C,CAC9E,CAAA;QAED,OAAO;YACL,aAAa,eAAA;YACb,gBAAgB,kBAAA;SACjB,CAAA;IAEH,CAAC;IAED;QACE,IAAM,aAAa,GAAG,0BAA0B,CAAC,IAAI,EAAE,GAAG,EAAE,CAAC,CAAC,CAAA;QAC9D,IAAM,aAAa,GAAG,0BAA0B,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC,CAAC,CAAA;QAC7D,IAAM,aAAa,GAAG,0BAA0B,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC,CAAC,CAAA;QAC7D,IAAM,aAAa,GAAG,0BAA0B,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC,CAAC,CAAA;QAC7D,IAAM,aAAa,GAAG,0BAA0B,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC,CAAC,CAAA;QAC7D,IAAM,aAAa,GAAG,0BAA0B,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC,CAAC,CAAA;QAC7D,IAAM,aAAa,GAAG,0BAA0B,CAAC,GAAG,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;QAC5D,IAAM,aAAa,GAAG,0BAA0B,CAAC,EAAE,EAAE,GAAG,EAAE,CAAC,CAAC,CAAA;QAE5D,IAAM,+BAA+B,GAAG,iBAAiB,CAAC,GAAG,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;QACrE,IAAM,wBAAwB,GAAG,iBAAiB,CAAC,GAAG,EAAE,CAAC,EAAE,CAAC,CAAC,CAAA;QAC7D,IAAM,+BAA+B,GAAG,iBAAiB,CAAC,IAAI,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;QACtE,IAAM,wBAAwB,GAAG,iBAAiB,CAAC,IAAI,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;QAC/D,IAAM,+BAA+B,GAAG,iBAAiB,CAAC,GAAG,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;QACrE,IAAM,wBAAwB,GAAG,iBAAiB,CAAC,GAAG,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;QAC9D,IAAM,+BAA+B,GAAG,iBAAiB,CAAC,GAAG,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;QACrE,IAAM,wBAAwB,GAAG,iBAAiB,CAAC,GAAG,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;QAC9D,IAAM,+BAA+B,GAAG,iBAAiB,CAAC,GAAG,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;QACrE,IAAM,wBAAwB,GAAG,iBAAiB,CAAC,GAAG,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;QAC9D,IAAM,+BAA+B,GAAG,iBAAiB,CAAC,GAAG,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;QACrE,IAAM,wBAAwB,GAAG,iBAAiB,CAAC,GAAG,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;QAE9D,IAAM,sBAAsB,GAAG;YAC7B,6BAA6B,EAAE,+BAA+B;YAC9D,sBAAsB,EAAE,wBAAwB;SACjD,CAAA;QACD,IAAM,sBAAsB,GAAG;YAC7B,6BAA6B,EAAE,+BAA+B;YAC9D,sBAAsB,EAAE,wBAAwB;SACjD,CAAA;QACD,IAAM,sBAAsB,GAAG;YAC7B,6BAA6B,EAAE,+BAA+B;YAC9D,sBAAsB,EAAE,wBAAwB;SACjD,CAAA;QACD,IAAM,sBAAsB,GAAG;YAC7B,6BAA6B,EAAE,+BAA+B;YAC9D,sBAAsB,EAAE,wBAAwB;SACjD,CAAA;QACD,IAAM,sBAAsB,GAAG;YAC7B,6BAA6B,EAAE,+BAA+B;YAC9D,sBAAsB,EAAE,wBAAwB;SACjD,CAAA;QACD,IAAM,sBAAsB,GAAG;YAC7B,6BAA6B,EAAE,+BAA+B;YAC9D,sBAAsB,EAAE,wBAAwB;SACjD,CAAA;QAED,OAAO;YACL,aAAa,eAAA;YACb,aAAa,eAAA;YACb,aAAa,eAAA;YACb,aAAa,eAAA;YACb,aAAa,eAAA;YACb,aAAa,eAAA;YACb,aAAa,eAAA;YACb,aAAa,eAAA;YACb,sBAAsB,wBAAA;YACtB,sBAAsB,wBAAA;YACtB,sBAAsB,wBAAA;YACtB,sBAAsB,wBAAA;YACtB,sBAAsB,wBAAA;YACtB,sBAAsB,wBAAA;SACvB,CAAA;IACH,CAAC;IAGD,OAAO;QACL,wBAAwB,0BAAA;QACxB,4BAA4B,8BAAA;KAC7B,CAAA;AAEH,CAAC;AAED,uBAA8B,OAAqB;IAC3C,IAAA,2DAG4B,EAFhC,kCAAc,EACd,4CAAmB,CACa;IAE5B,IAAA,sCAG+B,EAFnC,sDAAwB,EACxB,8DAA4B,CACO;IAErC,IAAM,kBAAkB,GAAG,wBAAwB,EAAE,CAAA;IACrD,IAAM,uBAAuB,GAAG,4BAA4B,EAAE,CAAA;IAC9D,IAAM,SAAS,GAAG,EAAE,CAAC,QAAQ,CAC3B,cAAc,CAAC,IAAI,GAAG,CAAC,CAAC,EACxB,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC,CAAC,CACb,CAAA;IACD,IAAM,mBAAmB,GAAG;QAC1B,SAAS,WAAA;KACV,CAAA;IAED,IAAI,mBAAmB,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE;QACtC,MAAM,IAAI,KAAK,CAAC,oCAAkC,mBAAmB,EAAE,CAAC,MAAQ,CAAC,CAAA;KAClF;IAED,OAAO;QACL,kBAAkB,oBAAA;QAClB,uBAAuB,yBAAA;QACvB,mBAAmB,qBAAA;KACpB,CAAA;AACH,CAAC;AA9BD,sCA8BC"} {"version":3,"file":"extractParams.js","sourceRoot":"","sources":["../../src/faceDetectionNet/extractParams.ts"],"names":[],"mappings":";;AAAA,0CAA4C;AAE5C,0EAAyE;AAIzE,2BAA2B,cAAsC,EAAE,aAA6B;IAE9F,oCAAoC,WAAmB,EAAE,YAAoB;QAE3E,IAAM,OAAO,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,CAAC,GAAG,CAAC,GAAG,WAAW,CAAC,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,WAAW,EAAE,CAAC,CAAC,CAAC,CAAA;QACxF,IAAM,gBAAgB,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,WAAW,CAAC,CAAC,CAAA;QACjE,IAAM,iBAAiB,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,WAAW,CAAC,CAAC,CAAA;QAClE,IAAM,eAAe,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,WAAW,CAAC,CAAC,CAAA;QAChE,IAAM,mBAAmB,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,WAAW,CAAC,CAAC,CAAA;QAEpE,aAAa,CAAC,IAAI,CAChB,EAAE,SAAS,EAAK,YAAY,aAAU,EAAE,EACxC,EAAE,SAAS,EAAK,YAAY,sBAAmB,EAAE,EACjD,EAAE,SAAS,EAAK,YAAY,uBAAoB,EAAE,EAClD,EAAE,SAAS,EAAK,YAAY,qBAAkB,EAAE,EAChD,EAAE,SAAS,EAAK,YAAY,yBAAsB,EAAE,CACrD,CAAA;QAED,OAAO;YACL,OAAO,SAAA;YACP,gBAAgB,kBAAA;YAChB,iBAAiB,mBAAA;YACjB,eAAe,iBAAA;YACf,mBAAmB,qBAAA;SACpB,CAAA;IACH,CAAC;IAED,2BACE,UAAkB,EAClB,WAAmB,EACnB,UAAkB,EAClB,YAAoB,EACpB,eAAyB;QAGzB,IAAM,OAAO,GAAG,EAAE,CAAC,QAAQ,CACzB,cAAc,CAAC,UAAU,GAAG,WAAW,GAAG,UAAU,GAAG,UAAU,CAAC,EAClE,CAAC,UAAU,EAAE,UAAU,EAAE,UAAU,EAAE,WAAW,CAAC,CAClD,CAAA;QACD,IAAM,IAAI,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,WAAW,CAAC,CAAC,CAAA;QAErD,aAAa,CAAC,IAAI,CAChB,EAAE,SAAS,EAAK,YAAY,aAAU,EAAE,EACxC,EAAE,SAAS,EAAK,YAAY,UAAI,eAAe,CAAC,CAAC,CAAC,mBAAmB,CAAC,CAAC,CAAC,MAAM,CAAE,EAAE,CACnF,CAAA;QAED,OAAO,EAAE,OAAO,SAAA,EAAE,IAAI,MAAA,EAAE,CAAA;IAC1B,CAAC;IAED,oCACE,UAAkB,EAClB,WAAmB,EACnB,UAAkB,EAClB,YAAoB;QAGd,IAAA,+EAGwE,EAF5E,oBAAO,EACP,cAAI,CACwE;QAE9E,OAAO;YACL,OAAO,SAAA;YACP,iBAAiB,EAAE,IAAI;SACxB,CAAA;IACH,CAAC;IAED,+BACE,UAAkB,EAClB,WAAmB,EACnB,YAAoB;QAGpB,IAAM,cAAc,GAAG,0BAA0B,CAAC,UAAU,EAAK,YAAY,oBAAiB,CAAC,CAAA;QAC/F,IAAM,cAAc,GAAG,0BAA0B,CAAC,UAAU,EAAE,WAAW,EAAE,CAAC,EAAK,YAAY,oBAAiB,CAAC,CAAA;QAE/G,OAAO,EAAE,cAAc,gBAAA,EAAE,cAAc,gBAAA,EAAE,CAAA;IAC3C,CAAC;IAED;QAEE,IAAM,MAAM,GAAG,0BAA0B,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,oBAAoB,CAAC,CAAA;QAEzE,IAAM,MAAM,GAAG,qBAAqB,CAAC,EAAE,EAAE,EAAE,EAAE,oBAAoB,CAAC,CAAA;QAClE,IAAM,MAAM,GAAG,qBAAqB,CAAC,EAAE,EAAE,GAAG,EAAE,oBAAoB,CAAC,CAAA;QACnE,IAAM,MAAM,GAAG,qBAAqB,CAAC,GAAG,EAAE,GAAG,EAAE,oBAAoB,CAAC,CAAA;QACpE,IAAM,MAAM,GAAG,qBAAqB,CAAC,GAAG,EAAE,GAAG,EAAE,oBAAoB,CAAC,CAAA;QACpE,IAAM,MAAM,GAAG,qBAAqB,CAAC,GAAG,EAAE,GAAG,EAAE,oBAAoB,CAAC,CAAA;QACpE,IAAM,MAAM,GAAG,qBAAqB,CAAC,GAAG,EAAE,GAAG,EAAE,oBAAoB,CAAC,CAAA;QACpE,IAAM,MAAM,GAAG,qBAAqB,CAAC,GAAG,EAAE,GAAG,EAAE,oBAAoB,CAAC,CAAA;QACpE,IAAM,MAAM,GAAG,qBAAqB,CAAC,GAAG,EAAE,GAAG,EAAE,oBAAoB,CAAC,CAAA;QACpE,IAAM,MAAM,GAAG,qBAAqB,CAAC,GAAG,EAAE,GAAG,EAAE,oBAAoB,CAAC,CAAA;QACpE,IAAM,OAAO,GAAG,qBAAqB,CAAC,GAAG,EAAE,GAAG,EAAE,qBAAqB,CAAC,CAAA;QACtE,IAAM,OAAO,GAAG,qBAAqB,CAAC,GAAG,EAAE,GAAG,EAAE,qBAAqB,CAAC,CAAA;QACtE,IAAM,OAAO,GAAG,qBAAqB,CAAC,GAAG,EAAE,IAAI,EAAE,qBAAqB,CAAC,CAAA;QACvE,IAAM,OAAO,GAAG,qBAAqB,CAAC,IAAI,EAAE,IAAI,EAAE,qBAAqB,CAAC,CAAA;QAExE,OAAO;YACL,MAAM,QAAA;YACN,MAAM,QAAA;YACN,MAAM,QAAA;YACN,MAAM,QAAA;YACN,MAAM,QAAA;YACN,MAAM,QAAA;YACN,MAAM,QAAA;YACN,MAAM,QAAA;YACN,MAAM,QAAA;YACN,MAAM,QAAA;YACN,OAAO,SAAA;YACP,OAAO,SAAA;YACP,OAAO,SAAA;YACP,OAAO,SAAA;SACR,CAAA;IACH,CAAC;IAED;QACE,IAAM,MAAM,GAAG,0BAA0B,CAAC,IAAI,EAAE,GAAG,EAAE,CAAC,EAAE,yBAAyB,CAAC,CAAA;QAClF,IAAM,MAAM,GAAG,0BAA0B,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC,EAAE,yBAAyB,CAAC,CAAA;QACjF,IAAM,MAAM,GAAG,0BAA0B,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC,EAAE,yBAAyB,CAAC,CAAA;QACjF,IAAM,MAAM,GAAG,0BAA0B,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC,EAAE,yBAAyB,CAAC,CAAA;QACjF,IAAM,MAAM,GAAG,0BAA0B,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC,EAAE,yBAAyB,CAAC,CAAA;QACjF,IAAM,MAAM,GAAG,0BAA0B,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC,EAAE,yBAAyB,CAAC,CAAA;QACjF,IAAM,MAAM,GAAG,0BAA0B,CAAC,GAAG,EAAE,EAAE,EAAE,CAAC,EAAE,yBAAyB,CAAC,CAAA;QAChF,IAAM,MAAM,GAAG,0BAA0B,CAAC,EAAE,EAAE,GAAG,EAAE,CAAC,EAAE,yBAAyB,CAAC,CAAA;QAEhF,IAAM,wBAAwB,GAAG,iBAAiB,CAAC,GAAG,EAAE,EAAE,EAAE,CAAC,EAAE,yDAAyD,CAAC,CAAA;QACzH,IAAM,iBAAiB,GAAG,iBAAiB,CAAC,GAAG,EAAE,CAAC,EAAE,CAAC,EAAE,kDAAkD,CAAC,CAAA;QAC1G,IAAM,wBAAwB,GAAG,iBAAiB,CAAC,IAAI,EAAE,EAAE,EAAE,CAAC,EAAE,yDAAyD,CAAC,CAAA;QAC1H,IAAM,iBAAiB,GAAG,iBAAiB,CAAC,IAAI,EAAE,EAAE,EAAE,CAAC,EAAE,kDAAkD,CAAC,CAAA;QAC5G,IAAM,wBAAwB,GAAG,iBAAiB,CAAC,GAAG,EAAE,EAAE,EAAE,CAAC,EAAE,yDAAyD,CAAC,CAAA;QACzH,IAAM,iBAAiB,GAAG,iBAAiB,CAAC,GAAG,EAAE,EAAE,EAAE,CAAC,EAAE,kDAAkD,CAAC,CAAA;QAC3G,IAAM,wBAAwB,GAAG,iBAAiB,CAAC,GAAG,EAAE,EAAE,EAAE,CAAC,EAAE,yDAAyD,CAAC,CAAA;QACzH,IAAM,iBAAiB,GAAG,iBAAiB,CAAC,GAAG,EAAE,EAAE,EAAE,CAAC,EAAE,kDAAkD,CAAC,CAAA;QAC3G,IAAM,wBAAwB,GAAG,iBAAiB,CAAC,GAAG,EAAE,EAAE,EAAE,CAAC,EAAE,yDAAyD,CAAC,CAAA;QACzH,IAAM,iBAAiB,GAAG,iBAAiB,CAAC,GAAG,EAAE,EAAE,EAAE,CAAC,EAAE,kDAAkD,CAAC,CAAA;QAC3G,IAAM,wBAAwB,GAAG,iBAAiB,CAAC,GAAG,EAAE,EAAE,EAAE,CAAC,EAAE,yDAAyD,CAAC,CAAA;QACzH,IAAM,iBAAiB,GAAG,iBAAiB,CAAC,GAAG,EAAE,EAAE,EAAE,CAAC,EAAE,kDAAkD,CAAC,CAAA;QAE3G,IAAM,eAAe,GAAG;YACtB,sBAAsB,EAAE,wBAAwB;YAChD,eAAe,EAAE,iBAAiB;SACnC,CAAA;QACD,IAAM,eAAe,GAAG;YACtB,sBAAsB,EAAE,wBAAwB;YAChD,eAAe,EAAE,iBAAiB;SACnC,CAAA;QACD,IAAM,eAAe,GAAG;YACtB,sBAAsB,EAAE,wBAAwB;YAChD,eAAe,EAAE,iBAAiB;SACnC,CAAA;QACD,IAAM,eAAe,GAAG;YACtB,sBAAsB,EAAE,wBAAwB;YAChD,eAAe,EAAE,iBAAiB;SACnC,CAAA;QACD,IAAM,eAAe,GAAG;YACtB,sBAAsB,EAAE,wBAAwB;YAChD,eAAe,EAAE,iBAAiB;SACnC,CAAA;QACD,IAAM,eAAe,GAAG;YACtB,sBAAsB,EAAE,wBAAwB;YAChD,eAAe,EAAE,iBAAiB;SACnC,CAAA;QAED,OAAO;YACL,MAAM,QAAA;YACN,MAAM,QAAA;YACN,MAAM,QAAA;YACN,MAAM,QAAA;YACN,MAAM,QAAA;YACN,MAAM,QAAA;YACN,MAAM,QAAA;YACN,MAAM,QAAA;YACN,eAAe,iBAAA;YACf,eAAe,iBAAA;YACf,eAAe,iBAAA;YACf,eAAe,iBAAA;YACf,eAAe,iBAAA;YACf,eAAe,iBAAA;SAChB,CAAA;IACH,CAAC;IAED,OAAO;QACL,wBAAwB,0BAAA;QACxB,4BAA4B,8BAAA;KAC7B,CAAA;AAEH,CAAC;AAED,uBAA8B,OAAqB;IAEjD,IAAM,aAAa,GAAmB,EAAE,CAAA;IAElC,IAAA,2DAG4B,EAFhC,kCAAc,EACd,4CAAmB,CACa;IAE5B,IAAA,qDAG8C,EAFlD,sDAAwB,EACxB,8DAA4B,CACsB;IAEpD,IAAM,WAAW,GAAG,wBAAwB,EAAE,CAAA;IAC9C,IAAM,gBAAgB,GAAG,4BAA4B,EAAE,CAAA;IACvD,IAAM,SAAS,GAAG,EAAE,CAAC,QAAQ,CAC3B,cAAc,CAAC,IAAI,GAAG,CAAC,CAAC,EACxB,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC,CAAC,CACb,CAAA;IACD,IAAM,YAAY,GAAG;QACnB,SAAS,WAAA;KACV,CAAA;IAED,aAAa,CAAC,IAAI,CAAC,EAAE,SAAS,EAAE,wBAAwB,EAAE,CAAC,CAAA;IAE3D,IAAI,mBAAmB,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE;QACtC,MAAM,IAAI,KAAK,CAAC,oCAAkC,mBAAmB,EAAE,CAAC,MAAQ,CAAC,CAAA;KAClF;IAED,OAAO;QACL,MAAM,EAAE;YACN,WAAW,aAAA;YACX,gBAAgB,kBAAA;YAChB,YAAY,cAAA;SACb;QACD,aAAa,eAAA;KACd,CAAA;AACH,CAAC;AAtCD,sCAsCC"}
\ No newline at end of file \ No newline at end of file
build/faceDetectionNet/loadQuantizedParams.d.ts
export declare function loadQuantizedParams(uri: string | undefined): Promise<any>; import { ParamMapping } from '../commons/types';
import { NetParams } from './types';
export declare function loadQuantizedParams(uri: string | undefined): Promise<{
params: NetParams;
paramMappings: ParamMapping[];
}>;
build/faceDetectionNet/loadQuantizedParams.js
"use strict"; "use strict";
Object.defineProperty(exports, "__esModule", { value: true }); Object.defineProperty(exports, "__esModule", { value: true });
var tslib_1 = require("tslib"); var tslib_1 = require("tslib");
var disposeUnusedWeightTensors_1 = require("../commons/disposeUnusedWeightTensors");
var extractWeightEntryFactory_1 = require("../commons/extractWeightEntryFactory");
var isTensor_1 = require("../commons/isTensor"); var isTensor_1 = require("../commons/isTensor");
var loadWeightMap_1 = require("../commons/loadWeightMap"); var loadWeightMap_1 = require("../commons/loadWeightMap");
var DEFAULT_MODEL_NAME = 'face_detection_model'; var DEFAULT_MODEL_NAME = 'face_detection_model';
function extractorsFactory(weightMap) { function extractorsFactory(weightMap, paramMappings) {
function extractPointwiseConvParams(prefix, idx) { var extractWeightEntry = extractWeightEntryFactory_1.extractWeightEntryFactory(weightMap, paramMappings);
var pointwise_conv_params = { function extractPointwiseConvParams(prefix, idx, mappedPrefix) {
filters: weightMap[prefix + "/Conv2d_" + idx + "_pointwise/weights"], var filters = extractWeightEntry(prefix + "/Conv2d_" + idx + "_pointwise/weights", 4, mappedPrefix + "/filters");
batch_norm_offset: weightMap[prefix + "/Conv2d_" + idx + "_pointwise/convolution_bn_offset"] var batch_norm_offset = extractWeightEntry(prefix + "/Conv2d_" + idx + "_pointwise/convolution_bn_offset", 1, mappedPrefix + "/batch_norm_offset");
}; return { filters: filters, batch_norm_offset: batch_norm_offset };
if (!isTensor_1.isTensor4D(pointwise_conv_params.filters)) {
throw new Error("expected weightMap[" + prefix + "/Conv2d_" + idx + "_pointwise/weights] to be a Tensor4D, instead have " + pointwise_conv_params.filters);
}
if (!isTensor_1.isTensor1D(pointwise_conv_params.batch_norm_offset)) {
throw new Error("expected weightMap[" + prefix + "/Conv2d_" + idx + "_pointwise/convolution_bn_offset] to be a Tensor1D, instead have " + pointwise_conv_params.batch_norm_offset);
}
return pointwise_conv_params;
} }
function extractConvPairParams(idx) { function extractConvPairParams(idx) {
var depthwise_conv_params = { var mappedPrefix = "mobilenetv1/conv_" + idx;
filters: weightMap["MobilenetV1/Conv2d_" + idx + "_depthwise/depthwise_weights"], var prefixDepthwiseConv = "MobilenetV1/Conv2d_" + idx + "_depthwise";
batch_norm_scale: weightMap["MobilenetV1/Conv2d_" + idx + "_depthwise/BatchNorm/gamma"], var mappedPrefixDepthwiseConv = mappedPrefix + "/depthwise_conv";
batch_norm_offset: weightMap["MobilenetV1/Conv2d_" + idx + "_depthwise/BatchNorm/beta"], var mappedPrefixPointwiseConv = mappedPrefix + "/pointwise_conv";
batch_norm_mean: weightMap["MobilenetV1/Conv2d_" + idx + "_depthwise/BatchNorm/moving_mean"], var filters = extractWeightEntry(prefixDepthwiseConv + "/depthwise_weights", 4, mappedPrefixDepthwiseConv + "/filters");
batch_norm_variance: weightMap["MobilenetV1/Conv2d_" + idx + "_depthwise/BatchNorm/moving_variance"], var batch_norm_scale = extractWeightEntry(prefixDepthwiseConv + "/BatchNorm/gamma", 1, mappedPrefixDepthwiseConv + "/batch_norm_scale");
}; var batch_norm_offset = extractWeightEntry(prefixDepthwiseConv + "/BatchNorm/beta", 1, mappedPrefixDepthwiseConv + "/batch_norm_offset");
if (!isTensor_1.isTensor4D(depthwise_conv_params.filters)) { var batch_norm_mean = extractWeightEntry(prefixDepthwiseConv + "/BatchNorm/moving_mean", 1, mappedPrefixDepthwiseConv + "/batch_norm_mean");
throw new Error("expected weightMap[MobilenetV1/Conv2d_" + idx + "_depthwise/depthwise_weights] to be a Tensor4D, instead have " + depthwise_conv_params.filters); var batch_norm_variance = extractWeightEntry(prefixDepthwiseConv + "/BatchNorm/moving_variance", 1, mappedPrefixDepthwiseConv + "/batch_norm_variance");
}
if (!isTensor_1.isTensor1D(depthwise_conv_params.batch_norm_scale)) {
throw new Error("expected weightMap[MobilenetV1/Conv2d_" + idx + "_depthwise/BatchNorm/gamma] to be a Tensor1D, instead have " + depthwise_conv_params.batch_norm_scale);
}
if (!isTensor_1.isTensor1D(depthwise_conv_params.batch_norm_offset)) {
throw new Error("expected weightMap[MobilenetV1/Conv2d_" + idx + "_depthwise/BatchNorm/beta] to be a Tensor1D, instead have " + depthwise_conv_params.batch_norm_offset);
}
if (!isTensor_1.isTensor1D(depthwise_conv_params.batch_norm_mean)) {
throw new Error("expected weightMap[MobilenetV1/Conv2d_" + idx + "_depthwise/BatchNorm/moving_mean] to be a Tensor1D, instead have " + depthwise_conv_params.batch_norm_mean);
}
if (!isTensor_1.isTensor1D(depthwise_conv_params.batch_norm_variance)) {
throw new Error("expected weightMap[MobilenetV1/Conv2d_" + idx + "_depthwise/BatchNorm/moving_variance] to be a Tensor1D, instead have " + depthwise_conv_params.batch_norm_variance);
}
return { return {
depthwise_conv_params: depthwise_conv_params, depthwise_conv: {
pointwise_conv_params: extractPointwiseConvParams('MobilenetV1', idx) filters: filters,
batch_norm_scale: batch_norm_scale,
batch_norm_offset: batch_norm_offset,
batch_norm_mean: batch_norm_mean,
batch_norm_variance: batch_norm_variance
},
pointwise_conv: extractPointwiseConvParams('MobilenetV1', idx, mappedPrefixPointwiseConv)
}; };
} }
function extractMobilenetV1Params() { function extractMobilenetV1Params() {
return { return {
conv_0_params: extractPointwiseConvParams('MobilenetV1', 0), conv_0: extractPointwiseConvParams('MobilenetV1', 0, 'mobilenetv1/conv_0'),
conv_pair_params: Array(13).fill(0).map(function (_, i) { return extractConvPairParams(i + 1); }) conv_1: extractConvPairParams(1),
conv_2: extractConvPairParams(2),
conv_3: extractConvPairParams(3),
conv_4: extractConvPairParams(4),
conv_5: extractConvPairParams(5),
conv_6: extractConvPairParams(6),
conv_7: extractConvPairParams(7),
conv_8: extractConvPairParams(8),
conv_9: extractConvPairParams(9),
conv_10: extractConvPairParams(10),
conv_11: extractConvPairParams(11),
conv_12: extractConvPairParams(12),
conv_13: extractConvPairParams(13)
}; };
} }
function extractBoxPredictorParams(idx) { function extractConvParams(prefix, mappedPrefix) {
var params = { var filters = extractWeightEntry(prefix + "/weights", 4, mappedPrefix + "/filters");
box_encoding_predictor_params: { var bias = extractWeightEntry(prefix + "/biases", 1, mappedPrefix + "/bias");
filters: weightMap["Prediction/BoxPredictor_" + idx + "/BoxEncodingPredictor/weights"], return { filters: filters, bias: bias };
bias: weightMap["Prediction/BoxPredictor_" + idx + "/BoxEncodingPredictor/biases"]
},
class_predictor_params: {
filters: weightMap["Prediction/BoxPredictor_" + idx + "/ClassPredictor/weights"],
bias: weightMap["Prediction/BoxPredictor_" + idx + "/ClassPredictor/biases"]
}
};
if (!isTensor_1.isTensor4D(params.box_encoding_predictor_params.filters)) {
throw new Error("expected weightMap[Prediction/BoxPredictor_" + idx + "/BoxEncodingPredictor/weights] to be a Tensor4D, instead have " + params.box_encoding_predictor_params.filters);
}
if (!isTensor_1.isTensor1D(params.box_encoding_predictor_params.bias)) {
throw new Error("expected weightMap[Prediction/BoxPredictor_" + idx + "/BoxEncodingPredictor/biases] to be a Tensor1D, instead have " + params.box_encoding_predictor_params.bias);
} }
if (!isTensor_1.isTensor4D(params.class_predictor_params.filters)) { function extractBoxPredictorParams(idx) {
throw new Error("expected weightMap[Prediction/BoxPredictor_" + idx + "/ClassPredictor/weights] to be a Tensor4D, instead have " + params.class_predictor_params.filters); var box_encoding_predictor = extractConvParams("Prediction/BoxPredictor_" + idx + "/BoxEncodingPredictor", "prediction_layer/box_predictor_" + idx + "/box_encoding_predictor");
} var class_predictor = extractConvParams("Prediction/BoxPredictor_" + idx + "/ClassPredictor", "prediction_layer/box_predictor_" + idx + "/class_predictor");
if (!isTensor_1.isTensor1D(params.class_predictor_params.bias)) { return { box_encoding_predictor: box_encoding_predictor, class_predictor: class_predictor };
throw new Error("expected weightMap[Prediction/BoxPredictor_" + idx + "/ClassPredictor/biases] to be a Tensor1D, instead have " + params.class_predictor_params.bias);
}
return params;
} }
function extractPredictionLayerParams() { function extractPredictionLayerParams() {
return { return {
conv_0_params: extractPointwiseConvParams('Prediction', 0), conv_0: extractPointwiseConvParams('Prediction', 0, 'prediction_layer/conv_0'),
conv_1_params: extractPointwiseConvParams('Prediction', 1), conv_1: extractPointwiseConvParams('Prediction', 1, 'prediction_layer/conv_1'),
conv_2_params: extractPointwiseConvParams('Prediction', 2), conv_2: extractPointwiseConvParams('Prediction', 2, 'prediction_layer/conv_2'),
conv_3_params: extractPointwiseConvParams('Prediction', 3), conv_3: extractPointwiseConvParams('Prediction', 3, 'prediction_layer/conv_3'),
conv_4_params: extractPointwiseConvParams('Prediction', 4), conv_4: extractPointwiseConvParams('Prediction', 4, 'prediction_layer/conv_4'),
conv_5_params: extractPointwiseConvParams('Prediction', 5), conv_5: extractPointwiseConvParams('Prediction', 5, 'prediction_layer/conv_5'),
conv_6_params: extractPointwiseConvParams('Prediction', 6), conv_6: extractPointwiseConvParams('Prediction', 6, 'prediction_layer/conv_6'),
conv_7_params: extractPointwiseConvParams('Prediction', 7), conv_7: extractPointwiseConvParams('Prediction', 7, 'prediction_layer/conv_7'),
box_predictor_0_params: extractBoxPredictorParams(0), box_predictor_0: extractBoxPredictorParams(0),
box_predictor_1_params: extractBoxPredictorParams(1), box_predictor_1: extractBoxPredictorParams(1),
box_predictor_2_params: extractBoxPredictorParams(2), box_predictor_2: extractBoxPredictorParams(2),
box_predictor_3_params: extractBoxPredictorParams(3), box_predictor_3: extractBoxPredictorParams(3),
box_predictor_4_params: extractBoxPredictorParams(4), box_predictor_4: extractBoxPredictorParams(4),
box_predictor_5_params: extractBoxPredictorParams(5) box_predictor_5: extractBoxPredictorParams(5)
}; };
} }
return { return {
...@@ -102,24 +87,28 @@ function extractorsFactory(weightMap) { ...@@ -102,24 +87,28 @@ function extractorsFactory(weightMap) {
} }
function loadQuantizedParams(uri) { function loadQuantizedParams(uri) {
return tslib_1.__awaiter(this, void 0, void 0, function () { return tslib_1.__awaiter(this, void 0, void 0, function () {
var weightMap, _a, extractMobilenetV1Params, extractPredictionLayerParams, extra_dim; var weightMap, paramMappings, _a, extractMobilenetV1Params, extractPredictionLayerParams, extra_dim, params;
return tslib_1.__generator(this, function (_b) { return tslib_1.__generator(this, function (_b) {
switch (_b.label) { switch (_b.label) {
case 0: return [4 /*yield*/, loadWeightMap_1.loadWeightMap(uri, DEFAULT_MODEL_NAME)]; case 0: return [4 /*yield*/, loadWeightMap_1.loadWeightMap(uri, DEFAULT_MODEL_NAME)];
case 1: case 1:
weightMap = _b.sent(); weightMap = _b.sent();
_a = extractorsFactory(weightMap), extractMobilenetV1Params = _a.extractMobilenetV1Params, extractPredictionLayerParams = _a.extractPredictionLayerParams; paramMappings = [];
_a = extractorsFactory(weightMap, paramMappings), extractMobilenetV1Params = _a.extractMobilenetV1Params, extractPredictionLayerParams = _a.extractPredictionLayerParams;
extra_dim = weightMap['Output/extra_dim']; extra_dim = weightMap['Output/extra_dim'];
paramMappings.push({ originalPath: 'Output/extra_dim', paramPath: 'output_layer/extra_dim' });
if (!isTensor_1.isTensor3D(extra_dim)) { if (!isTensor_1.isTensor3D(extra_dim)) {
throw new Error("expected weightMap['Output/extra_dim'] to be a Tensor3D, instead have " + extra_dim); throw new Error("expected weightMap['Output/extra_dim'] to be a Tensor3D, instead have " + extra_dim);
} }
return [2 /*return*/, { params = {
mobilenetv1_params: extractMobilenetV1Params(), mobilenetv1: extractMobilenetV1Params(),
prediction_layer_params: extractPredictionLayerParams(), prediction_layer: extractPredictionLayerParams(),
output_layer_params: { output_layer: {
extra_dim: extra_dim extra_dim: extra_dim
} }
}]; };
disposeUnusedWeightTensors_1.disposeUnusedWeightTensors(weightMap, paramMappings);
return [2 /*return*/, { params: params, paramMappings: paramMappings }];
} }
}); });
}); });
......
build/faceDetectionNet/loadQuantizedParams.js.map
{"version":3,"file":"loadQuantizedParams.js","sourceRoot":"","sources":["../../src/faceDetectionNet/loadQuantizedParams.ts"],"names":[],"mappings":";;;AAAA,gDAAyE;AACzE,0DAAyD;AAGzD,IAAM,kBAAkB,GAAG,sBAAsB,CAAA;AAEjD,2BAA2B,SAAc;IAEvC,oCAAoC,MAAc,EAAE,GAAW;QAE7D,IAAM,qBAAqB,GAAG;YAC5B,OAAO,EAAE,SAAS,CAAI,MAAM,gBAAW,GAAG,uBAAoB,CAAC;YAC/D,iBAAiB,EAAE,SAAS,CAAI,MAAM,gBAAW,GAAG,qCAAkC,CAAC;SACxF,CAAA;QAED,IAAI,CAAC,qBAAU,CAAC,qBAAqB,CAAC,OAAO,CAAC,EAAE;YAC9C,MAAM,IAAI,KAAK,CAAC,wBAAsB,MAAM,gBAAW,GAAG,2DAAsD,qBAAqB,CAAC,OAAS,CAAC,CAAA;SACjJ;QAED,IAAI,CAAC,qBAAU,CAAC,qBAAqB,CAAC,iBAAiB,CAAC,EAAE;YACxD,MAAM,IAAI,KAAK,CAAC,wBAAsB,MAAM,gBAAW,GAAG,yEAAoE,qBAAqB,CAAC,iBAAmB,CAAC,CAAA;SACzK;QAED,OAAO,qBAAqB,CAAA;IAC9B,CAAC;IAED,+BAA+B,GAAW;QAExC,IAAM,qBAAqB,GAAG;YAC5B,OAAO,EAAE,SAAS,CAAC,wBAAsB,GAAG,iCAA8B,CAAC;YAC3E,gBAAgB,EAAE,SAAS,CAAC,wBAAsB,GAAG,+BAA4B,CAAC;YAClF,iBAAiB,EAAE,SAAS,CAAC,wBAAsB,GAAG,8BAA2B,CAAC;YAClF,eAAe,EAAE,SAAS,CAAC,wBAAsB,GAAG,qCAAkC,CAAC;YACvF,mBAAmB,EAAE,SAAS,CAAC,wBAAsB,GAAG,yCAAsC,CAAC;SAChG,CAAA;QAED,IAAI,CAAC,qBAAU,CAAC,qBAAqB,CAAC,OAAO,CAAC,EAAE;YAC9C,MAAM,IAAI,KAAK,CAAC,2CAAyC,GAAG,qEAAgE,qBAAqB,CAAC,OAAS,CAAC,CAAA;SAC7J;QAED,IAAI,CAAC,qBAAU,CAAC,qBAAqB,CAAC,gBAAgB,CAAC,EAAE;YACvD,MAAM,IAAI,KAAK,CAAC,2CAAyC,GAAG,mEAA8D,qBAAqB,CAAC,gBAAkB,CAAC,CAAA;SACpK;QAED,IAAI,CAAC,qBAAU,CAAC,qBAAqB,CAAC,iBAAiB,CAAC,EAAE;YACxD,MAAM,IAAI,KAAK,CAAC,2CAAyC,GAAG,kEAA6D,qBAAqB,CAAC,iBAAmB,CAAC,CAAA;SACpK;QAED,IAAI,CAAC,qBAAU,CAAC,qBAAqB,CAAC,eAAe,CAAC,EAAE;YACtD,MAAM,IAAI,KAAK,CAAC,2CAAyC,GAAG,yEAAoE,qBAAqB,CAAC,eAAiB,CAAC,CAAA;SACzK;QAED,IAAI,CAAC,qBAAU,CAAC,qBAAqB,CAAC,mBAAmB,CAAC,EAAE;YAC1D,MAAM,IAAI,KAAK,CAAC,2CAAyC,GAAG,6EAAwE,qBAAqB,CAAC,mBAAqB,CAAC,CAAA;SACjL;QAED,OAAO;YACL,qBAAqB,uBAAA;YACrB,qBAAqB,EAAE,0BAA0B,CAAC,aAAa,EAAE,GAAG,CAAC;SACtE,CAAA;IACH,CAAC;IAED;QACE,OAAO;YACL,aAAa,EAAE,0BAA0B,CAAC,aAAa,EAAE,CAAC,CAAC;YAC3D,gBAAgB,EAAE,KAAK,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,UAAC,CAAC,EAAE,CAAC,IAAK,OAAA,qBAAqB,CAAC,CAAC,GAAG,CAAC,CAAC,EAA5B,CAA4B,CAAC;SAChF,CAAA;IACH,CAAC;IAED,mCAAmC,GAAW;QAE5C,IAAM,MAAM,GAAG;YACb,6BAA6B,EAAE;gBAC7B,OAAO,EAAE,SAAS,CAAC,6BAA2B,GAAG,kCAA+B,CAAC;gBACjF,IAAI,EAAE,SAAS,CAAC,6BAA2B,GAAG,iCAA8B,CAAC;aAC9E;YACD,sBAAsB,EAAE;gBACtB,OAAO,EAAE,SAAS,CAAC,6BAA2B,GAAG,4BAAyB,CAAC;gBAC3E,IAAI,EAAE,SAAS,CAAC,6BAA2B,GAAG,2BAAwB,CAAC;aACxE;SACF,CAAA;QAED,IAAI,CAAC,qBAAU,CAAC,MAAM,CAAC,6BAA6B,CAAC,OAAO,CAAC,EAAE;YAC7D,MAAM,IAAI,KAAK,CAAC,gDAA8C,GAAG,sEAAiE,MAAM,CAAC,6BAA6B,CAAC,OAAS,CAAC,CAAA;SAClL;QAED,IAAI,CAAC,qBAAU,CAAC,MAAM,CAAC,6BAA6B,CAAC,IAAI,CAAC,EAAE;YAC1D,MAAM,IAAI,KAAK,CAAC,gDAA8C,GAAG,qEAAgE,MAAM,CAAC,6BAA6B,CAAC,IAAM,CAAC,CAAA;SAC9K;QAED,IAAI,CAAC,qBAAU,CAAC,MAAM,CAAC,sBAAsB,CAAC,OAAO,CAAC,EAAE;YACtD,MAAM,IAAI,KAAK,CAAC,gDAA8C,GAAG,gEAA2D,MAAM,CAAC,sBAAsB,CAAC,OAAS,CAAC,CAAA;SACrK;QAED,IAAI,CAAC,qBAAU,CAAC,MAAM,CAAC,sBAAsB,CAAC,IAAI,CAAC,EAAE;YACnD,MAAM,IAAI,KAAK,CAAC,gDAA8C,GAAG,+DAA0D,MAAM,CAAC,sBAAsB,CAAC,IAAM,CAAC,CAAA;SACjK;QAED,OAAO,MAAM,CAAA;IACf,CAAC;IAED;QACE,OAAO;YACL,aAAa,EAAE,0BAA0B,CAAC,YAAY,EAAE,CAAC,CAAC;YAC1D,aAAa,EAAE,0BAA0B,CAAC,YAAY,EAAE,CAAC,CAAC;YAC1D,aAAa,EAAE,0BAA0B,CAAC,YAAY,EAAE,CAAC,CAAC;YAC1D,aAAa,EAAE,0BAA0B,CAAC,YAAY,EAAE,CAAC,CAAC;YAC1D,aAAa,EAAE,0BAA0B,CAAC,YAAY,EAAE,CAAC,CAAC;YAC1D,aAAa,EAAE,0BAA0B,CAAC,YAAY,EAAE,CAAC,CAAC;YAC1D,aAAa,EAAE,0BAA0B,CAAC,YAAY,EAAE,CAAC,CAAC;YAC1D,aAAa,EAAE,0BAA0B,CAAC,YAAY,EAAE,CAAC,CAAC;YAC1D,sBAAsB,EAAE,yBAAyB,CAAC,CAAC,CAAC;YACpD,sBAAsB,EAAE,yBAAyB,CAAC,CAAC,CAAC;YACpD,sBAAsB,EAAE,yBAAyB,CAAC,CAAC,CAAC;YACpD,sBAAsB,EAAE,yBAAyB,CAAC,CAAC,CAAC;YACpD,sBAAsB,EAAE,yBAAyB,CAAC,CAAC,CAAC;YACpD,sBAAsB,EAAE,yBAAyB,CAAC,CAAC,CAAC;SACrD,CAAA;IACH,CAAC;IAED,OAAO;QACL,wBAAwB,0BAAA;QACxB,4BAA4B,8BAAA;KAC7B,CAAA;AACH,CAAC;AAED,6BAA0C,GAAuB;;;;;wBAC7C,qBAAM,6BAAa,CAAC,GAAG,EAAE,kBAAkB,CAAC,EAAA;;oBAAxD,SAAS,GAAG,SAA4C;oBAExD,KAGF,iBAAiB,CAAC,SAAS,CAAC,EAF9B,wBAAwB,8BAAA,EACxB,4BAA4B,kCAAA,CACE;oBAE1B,SAAS,GAAG,SAAS,CAAC,kBAAkB,CAAC,CAAA;oBAC/C,IAAI,CAAC,qBAAU,CAAC,SAAS,CAAC,EAAE;wBAC1B,MAAM,IAAI,KAAK,CAAC,2EAAyE,SAAW,CAAC,CAAA;qBACtG;oBAED,sBAAO;4BACL,kBAAkB,EAAE,wBAAwB,EAAE;4BAC9C,uBAAuB,EAAE,4BAA4B,EAAE;4BACvD,mBAAmB,EAAE;gCACnB,SAAS,WAAA;6BACV;yBACF,EAAA;;;;CACF;AApBD,kDAoBC"} {"version":3,"file":"loadQuantizedParams.js","sourceRoot":"","sources":["../../src/faceDetectionNet/loadQuantizedParams.ts"],"names":[],"mappings":";;;AACA,oFAAmF;AACnF,kFAAiF;AACjF,gDAAyE;AACzE,0DAAyD;AAIzD,IAAM,kBAAkB,GAAG,sBAAsB,CAAA;AAEjD,2BAA2B,SAAc,EAAE,aAA6B;IAEtE,IAAM,kBAAkB,GAAG,qDAAyB,CAAC,SAAS,EAAE,aAAa,CAAC,CAAA;IAE9E,oCAAoC,MAAc,EAAE,GAAW,EAAE,YAAoB;QAEnF,IAAM,OAAO,GAAG,kBAAkB,CAAiB,MAAM,gBAAW,GAAG,uBAAoB,EAAE,CAAC,EAAK,YAAY,aAAU,CAAC,CAAA;QAC1H,IAAM,iBAAiB,GAAG,kBAAkB,CAAiB,MAAM,gBAAW,GAAG,qCAAkC,EAAE,CAAC,EAAK,YAAY,uBAAoB,CAAC,CAAA;QAE5J,OAAO,EAAE,OAAO,SAAA,EAAE,iBAAiB,mBAAA,EAAE,CAAA;IACvC,CAAC;IAED,+BAA+B,GAAW;QAExC,IAAM,YAAY,GAAG,sBAAoB,GAAK,CAAA;QAC9C,IAAM,mBAAmB,GAAG,wBAAsB,GAAG,eAAY,CAAA;QACjE,IAAM,yBAAyB,GAAM,YAAY,oBAAiB,CAAA;QAClE,IAAM,yBAAyB,GAAM,YAAY,oBAAiB,CAAA;QAElE,IAAM,OAAO,GAAG,kBAAkB,CAAiB,mBAAmB,uBAAoB,EAAE,CAAC,EAAK,yBAAyB,aAAU,CAAC,CAAA;QACtI,IAAM,gBAAgB,GAAG,kBAAkB,CAAiB,mBAAmB,qBAAkB,EAAE,CAAC,EAAK,yBAAyB,sBAAmB,CAAC,CAAA;QACtJ,IAAM,iBAAiB,GAAG,kBAAkB,CAAiB,mBAAmB,oBAAiB,EAAE,CAAC,EAAK,yBAAyB,uBAAoB,CAAC,CAAA;QACvJ,IAAM,eAAe,GAAG,kBAAkB,CAAiB,mBAAmB,2BAAwB,EAAE,CAAC,EAAK,yBAAyB,qBAAkB,CAAC,CAAA;QAC1J,IAAM,mBAAmB,GAAG,kBAAkB,CAAiB,mBAAmB,+BAA4B,EAAE,CAAC,EAAK,yBAAyB,yBAAsB,CAAC,CAAA;QAEtK,OAAO;YACL,cAAc,EAAE;gBACd,OAAO,SAAA;gBACP,gBAAgB,kBAAA;gBAChB,iBAAiB,mBAAA;gBACjB,eAAe,iBAAA;gBACf,mBAAmB,qBAAA;aACpB;YACD,cAAc,EAAE,0BAA0B,CAAC,aAAa,EAAE,GAAG,EAAE,yBAAyB,CAAC;SAC1F,CAAA;IACH,CAAC;IAED;QACE,OAAO;YACL,MAAM,EAAE,0BAA0B,CAAC,aAAa,EAAE,CAAC,EAAE,oBAAoB,CAAC;YAC1E,MAAM,EAAE,qBAAqB,CAAC,CAAC,CAAC;YAChC,MAAM,EAAE,qBAAqB,CAAC,CAAC,CAAC;YAChC,MAAM,EAAE,qBAAqB,CAAC,CAAC,CAAC;YAChC,MAAM,EAAE,qBAAqB,CAAC,CAAC,CAAC;YAChC,MAAM,EAAE,qBAAqB,CAAC,CAAC,CAAC;YAChC,MAAM,EAAE,qBAAqB,CAAC,CAAC,CAAC;YAChC,MAAM,EAAE,qBAAqB,CAAC,CAAC,CAAC;YAChC,MAAM,EAAE,qBAAqB,CAAC,CAAC,CAAC;YAChC,MAAM,EAAE,qBAAqB,CAAC,CAAC,CAAC;YAChC,OAAO,EAAE,qBAAqB,CAAC,EAAE,CAAC;YAClC,OAAO,EAAE,qBAAqB,CAAC,EAAE,CAAC;YAClC,OAAO,EAAE,qBAAqB,CAAC,EAAE,CAAC;YAClC,OAAO,EAAE,qBAAqB,CAAC,EAAE,CAAC;SACnC,CAAA;IACH,CAAC;IAED,2BAA2B,MAAc,EAAE,YAAoB;QAC7D,IAAM,OAAO,GAAG,kBAAkB,CAAiB,MAAM,aAAU,EAAE,CAAC,EAAK,YAAY,aAAU,CAAC,CAAA;QAClG,IAAM,IAAI,GAAG,kBAAkB,CAAiB,MAAM,YAAS,EAAE,CAAC,EAAK,YAAY,UAAO,CAAC,CAAA;QAE3F,OAAO,EAAE,OAAO,SAAA,EAAE,IAAI,MAAA,EAAE,CAAA;IAC1B,CAAC;IAED,mCAAmC,GAAW;QAE5C,IAAM,sBAAsB,GAAG,iBAAiB,CAC9C,6BAA2B,GAAG,0BAAuB,EACrD,oCAAkC,GAAG,4BAAyB,CAC/D,CAAA;QACD,IAAM,eAAe,GAAG,iBAAiB,CACvC,6BAA2B,GAAG,oBAAiB,EAC/C,oCAAkC,GAAG,qBAAkB,CACxD,CAAA;QAED,OAAO,EAAE,sBAAsB,wBAAA,EAAE,eAAe,iBAAA,EAAE,CAAA;IACpD,CAAC;IAED;QACE,OAAO;YACL,MAAM,EAAE,0BAA0B,CAAC,YAAY,EAAE,CAAC,EAAE,yBAAyB,CAAC;YAC9E,MAAM,EAAE,0BAA0B,CAAC,YAAY,EAAE,CAAC,EAAE,yBAAyB,CAAC;YAC9E,MAAM,EAAE,0BAA0B,CAAC,YAAY,EAAE,CAAC,EAAE,yBAAyB,CAAC;YAC9E,MAAM,EAAE,0BAA0B,CAAC,YAAY,EAAE,CAAC,EAAE,yBAAyB,CAAC;YAC9E,MAAM,EAAE,0BAA0B,CAAC,YAAY,EAAE,CAAC,EAAE,yBAAyB,CAAC;YAC9E,MAAM,EAAE,0BAA0B,CAAC,YAAY,EAAE,CAAC,EAAE,yBAAyB,CAAC;YAC9E,MAAM,EAAE,0BAA0B,CAAC,YAAY,EAAE,CAAC,EAAE,yBAAyB,CAAC;YAC9E,MAAM,EAAE,0BAA0B,CAAC,YAAY,EAAE,CAAC,EAAE,yBAAyB,CAAC;YAC9E,eAAe,EAAE,yBAAyB,CAAC,CAAC,CAAC;YAC7C,eAAe,EAAE,yBAAyB,CAAC,CAAC,CAAC;YAC7C,eAAe,EAAE,yBAAyB,CAAC,CAAC,CAAC;YAC7C,eAAe,EAAE,yBAAyB,CAAC,CAAC,CAAC;YAC7C,eAAe,EAAE,yBAAyB,CAAC,CAAC,CAAC;YAC7C,eAAe,EAAE,yBAAyB,CAAC,CAAC,CAAC;SAC9C,CAAA;IACH,CAAC;IAED,OAAO;QACL,wBAAwB,0BAAA;QACxB,4BAA4B,8BAAA;KAC7B,CAAA;AACH,CAAC;AAED,6BACE,GAAuB;;;;;wBAGL,qBAAM,6BAAa,CAAC,GAAG,EAAE,kBAAkB,CAAC,EAAA;;oBAAxD,SAAS,GAAG,SAA4C;oBACxD,aAAa,GAAmB,EAAE,CAAA;oBAElC,KAGF,iBAAiB,CAAC,SAAS,EAAE,aAAa,CAAC,EAF7C,wBAAwB,8BAAA,EACxB,4BAA4B,kCAAA,CACiB;oBAEzC,SAAS,GAAG,SAAS,CAAC,kBAAkB,CAAC,CAAA;oBAC/C,aAAa,CAAC,IAAI,CAAC,EAAE,YAAY,EAAE,kBAAkB,EAAE,SAAS,EAAE,wBAAwB,EAAE,CAAC,CAAA;oBAE7F,IAAI,CAAC,qBAAU,CAAC,SAAS,CAAC,EAAE;wBAC1B,MAAM,IAAI,KAAK,CAAC,2EAAyE,SAAW,CAAC,CAAA;qBACtG;oBAEK,MAAM,GAAG;wBACb,WAAW,EAAE,wBAAwB,EAAE;wBACvC,gBAAgB,EAAE,4BAA4B,EAAE;wBAChD,YAAY,EAAE;4BACZ,SAAS,WAAA;yBACV;qBACF,CAAA;oBAED,uDAA0B,CAAC,SAAS,EAAE,aAAa,CAAC,CAAA;oBAEpD,sBAAO,EAAE,MAAM,QAAA,EAAE,aAAa,eAAA,EAAE,EAAA;;;;CACjC;AA9BD,kDA8BC"}
\ No newline at end of file \ No newline at end of file
build/faceDetectionNet/mobileNetV1.js
...@@ -16,12 +16,27 @@ function getStridesForLayerIdx(layerIdx) { ...@@ -16,12 +16,27 @@ function getStridesForLayerIdx(layerIdx) {
function mobileNetV1(x, params) { function mobileNetV1(x, params) {
return tf.tidy(function () { return tf.tidy(function () {
var conv11 = null; var conv11 = null;
var out = pointwiseConvLayer_1.pointwiseConvLayer(x, params.conv_0_params, [2, 2]); var out = pointwiseConvLayer_1.pointwiseConvLayer(x, params.conv_0, [2, 2]);
params.conv_pair_params.forEach(function (param, i) { var convPairParams = [
params.conv_1,
params.conv_2,
params.conv_3,
params.conv_4,
params.conv_5,
params.conv_6,
params.conv_7,
params.conv_8,
params.conv_9,
params.conv_10,
params.conv_11,
params.conv_12,
params.conv_13
];
convPairParams.forEach(function (param, i) {
var layerIdx = i + 1; var layerIdx = i + 1;
var depthwiseConvStrides = getStridesForLayerIdx(layerIdx); var depthwiseConvStrides = getStridesForLayerIdx(layerIdx);
out = depthwiseConvLayer(out, param.depthwise_conv_params, depthwiseConvStrides); out = depthwiseConvLayer(out, param.depthwise_conv, depthwiseConvStrides);
out = pointwiseConvLayer_1.pointwiseConvLayer(out, param.pointwise_conv_params, [1, 1]); out = pointwiseConvLayer_1.pointwiseConvLayer(out, param.pointwise_conv, [1, 1]);
if (layerIdx === 11) { if (layerIdx === 11) {
conv11 = out; conv11 = out;
} }
......
build/faceDetectionNet/mobileNetV1.js.map
{"version":3,"file":"mobileNetV1.js","sourceRoot":"","sources":["../../src/faceDetectionNet/mobileNetV1.ts"],"names":[],"mappings":";;AAAA,0CAA4C;AAE5C,2DAA0D;AAG1D,IAAM,OAAO,GAAG,qBAAqB,CAAA;AAErC,4BACE,CAAc,EACd,MAAuC,EACvC,OAAyB;IAEzB,OAAO,EAAE,CAAC,IAAI,CAAC;QAEb,IAAI,GAAG,GAAG,EAAE,CAAC,eAAe,CAAC,CAAC,EAAE,MAAM,CAAC,OAAO,EAAE,OAAO,EAAE,MAAM,CAAC,CAAA;QAChE,GAAG,GAAG,EAAE,CAAC,kBAAkB,CACzB,GAAG,EACH,MAAM,CAAC,eAAe,EACtB,MAAM,CAAC,mBAAmB,EAC1B,OAAO,EACP,MAAM,CAAC,gBAAgB,EACvB,MAAM,CAAC,iBAAiB,CACzB,CAAA;QACD,OAAO,EAAE,CAAC,WAAW,CAAC,GAAG,EAAE,CAAC,EAAE,CAAC,CAAC,CAAA;IAElC,CAAC,CAAC,CAAA;AACJ,CAAC;AAED,+BAA+B,QAAgB;IAC7C,OAAO,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,UAAA,GAAG,IAAI,OAAA,GAAG,KAAK,QAAQ,EAAhB,CAAgB,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAA;AACtE,CAAC;AAED,qBAA4B,CAAc,EAAE,MAA0B;IACpE,OAAO,EAAE,CAAC,IAAI,CAAC;QAEb,IAAI,MAAM,GAAG,IAAI,CAAA;QACjB,IAAI,GAAG,GAAG,uCAAkB,CAAC,CAAC,EAAE,MAAM,CAAC,aAAa,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QAE7D,MAAM,CAAC,gBAAgB,CAAC,OAAO,CAAC,UAAC,KAAK,EAAE,CAAC;YACvC,IAAM,QAAQ,GAAG,CAAC,GAAG,CAAC,CAAA;YACtB,IAAM,oBAAoB,GAAG,qBAAqB,CAAC,QAAQ,CAAC,CAAA;YAC5D,GAAG,GAAG,kBAAkB,CAAC,GAAG,EAAE,KAAK,CAAC,qBAAqB,EAAE,oBAAoB,CAAC,CAAA;YAChF,GAAG,GAAG,uCAAkB,CAAC,GAAG,EAAE,KAAK,CAAC,qBAAqB,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;YAClE,IAAI,QAAQ,KAAK,EAAE,EAAE;gBACnB,MAAM,GAAG,GAAG,CAAA;aACb;QACH,CAAC,CAAC,CAAA;QAEF,IAAI,MAAM,KAAK,IAAI,EAAE;YACnB,MAAM,IAAI,KAAK,CAAC,+CAA+C,CAAC,CAAA;SACjE;QAED,OAAO;YACL,GAAG,KAAA;YACH,MAAM,EAAE,MAAa;SACtB,CAAA;IAEH,CAAC,CAAC,CAAA;AACJ,CAAC;AA1BD,kCA0BC"} {"version":3,"file":"mobileNetV1.js","sourceRoot":"","sources":["../../src/faceDetectionNet/mobileNetV1.ts"],"names":[],"mappings":";;AAAA,0CAA4C;AAE5C,2DAA0D;AAG1D,IAAM,OAAO,GAAG,qBAAqB,CAAA;AAErC,4BACE,CAAc,EACd,MAAuC,EACvC,OAAyB;IAEzB,OAAO,EAAE,CAAC,IAAI,CAAC;QAEb,IAAI,GAAG,GAAG,EAAE,CAAC,eAAe,CAAC,CAAC,EAAE,MAAM,CAAC,OAAO,EAAE,OAAO,EAAE,MAAM,CAAC,CAAA;QAChE,GAAG,GAAG,EAAE,CAAC,kBAAkB,CACzB,GAAG,EACH,MAAM,CAAC,eAAe,EACtB,MAAM,CAAC,mBAAmB,EAC1B,OAAO,EACP,MAAM,CAAC,gBAAgB,EACvB,MAAM,CAAC,iBAAiB,CACzB,CAAA;QACD,OAAO,EAAE,CAAC,WAAW,CAAC,GAAG,EAAE,CAAC,EAAE,CAAC,CAAC,CAAA;IAElC,CAAC,CAAC,CAAA;AACJ,CAAC;AAED,+BAA+B,QAAgB;IAC7C,OAAO,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,UAAA,GAAG,IAAI,OAAA,GAAG,KAAK,QAAQ,EAAhB,CAAgB,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAA;AACtE,CAAC;AAED,qBAA4B,CAAc,EAAE,MAA0B;IACpE,OAAO,EAAE,CAAC,IAAI,CAAC;QAEb,IAAI,MAAM,GAAG,IAAI,CAAA;QACjB,IAAI,GAAG,GAAG,uCAAkB,CAAC,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QAEtD,IAAM,cAAc,GAAG;YACrB,MAAM,CAAC,MAAM;YACb,MAAM,CAAC,MAAM;YACb,MAAM,CAAC,MAAM;YACb,MAAM,CAAC,MAAM;YACb,MAAM,CAAC,MAAM;YACb,MAAM,CAAC,MAAM;YACb,MAAM,CAAC,MAAM;YACb,MAAM,CAAC,MAAM;YACb,MAAM,CAAC,MAAM;YACb,MAAM,CAAC,OAAO;YACd,MAAM,CAAC,OAAO;YACd,MAAM,CAAC,OAAO;YACd,MAAM,CAAC,OAAO;SACf,CAAA;QAED,cAAc,CAAC,OAAO,CAAC,UAAC,KAAK,EAAE,CAAC;YAC9B,IAAM,QAAQ,GAAG,CAAC,GAAG,CAAC,CAAA;YACtB,IAAM,oBAAoB,GAAG,qBAAqB,CAAC,QAAQ,CAAC,CAAA;YAC5D,GAAG,GAAG,kBAAkB,CAAC,GAAG,EAAE,KAAK,CAAC,cAAc,EAAE,oBAAoB,CAAC,CAAA;YACzE,GAAG,GAAG,uCAAkB,CAAC,GAAG,EAAE,KAAK,CAAC,cAAc,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;YAC3D,IAAI,QAAQ,KAAK,EAAE,EAAE;gBACnB,MAAM,GAAG,GAAG,CAAA;aACb;QACH,CAAC,CAAC,CAAA;QAEF,IAAI,MAAM,KAAK,IAAI,EAAE;YACnB,MAAM,IAAI,KAAK,CAAC,+CAA+C,CAAC,CAAA;SACjE;QAED,OAAO;YACL,GAAG,KAAA;YACH,MAAM,EAAE,MAAa;SACtB,CAAA;IAEH,CAAC,CAAC,CAAA;AACJ,CAAC;AA1CD,kCA0CC"}
\ No newline at end of file \ No newline at end of file
build/faceDetectionNet/predictionLayer.js
...@@ -5,20 +5,20 @@ var boxPredictionLayer_1 = require("./boxPredictionLayer"); ...@@ -5,20 +5,20 @@ var boxPredictionLayer_1 = require("./boxPredictionLayer");
var pointwiseConvLayer_1 = require("./pointwiseConvLayer"); var pointwiseConvLayer_1 = require("./pointwiseConvLayer");
function predictionLayer(x, conv11, params) { function predictionLayer(x, conv11, params) {
return tf.tidy(function () { return tf.tidy(function () {
var conv0 = pointwiseConvLayer_1.pointwiseConvLayer(x, params.conv_0_params, [1, 1]); var conv0 = pointwiseConvLayer_1.pointwiseConvLayer(x, params.conv_0, [1, 1]);
var conv1 = pointwiseConvLayer_1.pointwiseConvLayer(conv0, params.conv_1_params, [2, 2]); var conv1 = pointwiseConvLayer_1.pointwiseConvLayer(conv0, params.conv_1, [2, 2]);
var conv2 = pointwiseConvLayer_1.pointwiseConvLayer(conv1, params.conv_2_params, [1, 1]); var conv2 = pointwiseConvLayer_1.pointwiseConvLayer(conv1, params.conv_2, [1, 1]);
var conv3 = pointwiseConvLayer_1.pointwiseConvLayer(conv2, params.conv_3_params, [2, 2]); var conv3 = pointwiseConvLayer_1.pointwiseConvLayer(conv2, params.conv_3, [2, 2]);
var conv4 = pointwiseConvLayer_1.pointwiseConvLayer(conv3, params.conv_4_params, [1, 1]); var conv4 = pointwiseConvLayer_1.pointwiseConvLayer(conv3, params.conv_4, [1, 1]);
var conv5 = pointwiseConvLayer_1.pointwiseConvLayer(conv4, params.conv_5_params, [2, 2]); var conv5 = pointwiseConvLayer_1.pointwiseConvLayer(conv4, params.conv_5, [2, 2]);
var conv6 = pointwiseConvLayer_1.pointwiseConvLayer(conv5, params.conv_6_params, [1, 1]); var conv6 = pointwiseConvLayer_1.pointwiseConvLayer(conv5, params.conv_6, [1, 1]);
var conv7 = pointwiseConvLayer_1.pointwiseConvLayer(conv6, params.conv_7_params, [2, 2]); var conv7 = pointwiseConvLayer_1.pointwiseConvLayer(conv6, params.conv_7, [2, 2]);
var boxPrediction0 = boxPredictionLayer_1.boxPredictionLayer(conv11, params.box_predictor_0_params); var boxPrediction0 = boxPredictionLayer_1.boxPredictionLayer(conv11, params.box_predictor_0);
var boxPrediction1 = boxPredictionLayer_1.boxPredictionLayer(x, params.box_predictor_1_params); var boxPrediction1 = boxPredictionLayer_1.boxPredictionLayer(x, params.box_predictor_1);
var boxPrediction2 = boxPredictionLayer_1.boxPredictionLayer(conv1, params.box_predictor_2_params); var boxPrediction2 = boxPredictionLayer_1.boxPredictionLayer(conv1, params.box_predictor_2);
var boxPrediction3 = boxPredictionLayer_1.boxPredictionLayer(conv3, params.box_predictor_3_params); var boxPrediction3 = boxPredictionLayer_1.boxPredictionLayer(conv3, params.box_predictor_3);
var boxPrediction4 = boxPredictionLayer_1.boxPredictionLayer(conv5, params.box_predictor_4_params); var boxPrediction4 = boxPredictionLayer_1.boxPredictionLayer(conv5, params.box_predictor_4);
var boxPrediction5 = boxPredictionLayer_1.boxPredictionLayer(conv7, params.box_predictor_5_params); var boxPrediction5 = boxPredictionLayer_1.boxPredictionLayer(conv7, params.box_predictor_5);
var boxPredictions = tf.concat([ var boxPredictions = tf.concat([
boxPrediction0.boxPredictionEncoding, boxPrediction0.boxPredictionEncoding,
boxPrediction1.boxPredictionEncoding, boxPrediction1.boxPredictionEncoding,
......
build/faceDetectionNet/predictionLayer.js.map
{"version":3,"file":"predictionLayer.js","sourceRoot":"","sources":["../../src/faceDetectionNet/predictionLayer.ts"],"names":[],"mappings":";;AAAA,0CAA4C;AAE5C,2DAA0D;AAC1D,2DAA0D;AAG1D,yBACE,CAAc,EACd,MAAmB,EACnB,MAA6B;IAE7B,OAAO,EAAE,CAAC,IAAI,CAAC;QAEb,IAAM,KAAK,GAAG,uCAAkB,CAAC,CAAC,EAAE,MAAM,CAAC,aAAa,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QACjE,IAAM,KAAK,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,aAAa,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QACrE,IAAM,KAAK,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,aAAa,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QACrE,IAAM,KAAK,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,aAAa,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QACrE,IAAM,KAAK,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,aAAa,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QACrE,IAAM,KAAK,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,aAAa,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QACrE,IAAM,KAAK,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,aAAa,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QACrE,IAAM,KAAK,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,aAAa,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QAErE,IAAM,cAAc,GAAG,uCAAkB,CAAC,MAAM,EAAE,MAAM,CAAC,sBAAsB,CAAC,CAAA;QAChF,IAAM,cAAc,GAAG,uCAAkB,CAAC,CAAC,EAAE,MAAM,CAAC,sBAAsB,CAAC,CAAA;QAC3E,IAAM,cAAc,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,sBAAsB,CAAC,CAAA;QAC/E,IAAM,cAAc,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,sBAAsB,CAAC,CAAA;QAC/E,IAAM,cAAc,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,sBAAsB,CAAC,CAAA;QAC/E,IAAM,cAAc,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,sBAAsB,CAAC,CAAA;QAE/E,IAAM,cAAc,GAAG,EAAE,CAAC,MAAM,CAAC;YAC/B,cAAc,CAAC,qBAAqB;YACpC,cAAc,CAAC,qBAAqB;YACpC,cAAc,CAAC,qBAAqB;YACpC,cAAc,CAAC,qBAAqB;YACpC,cAAc,CAAC,qBAAqB;YACpC,cAAc,CAAC,qBAAqB;SACrC,EAAE,CAAC,CAAgB,CAAA;QAEpB,IAAM,gBAAgB,GAAG,EAAE,CAAC,MAAM,CAAC;YACjC,cAAc,CAAC,eAAe;YAC9B,cAAc,CAAC,eAAe;YAC9B,cAAc,CAAC,eAAe;YAC9B,cAAc,CAAC,eAAe;YAC9B,cAAc,CAAC,eAAe;YAC9B,cAAc,CAAC,eAAe;SAC/B,EAAE,CAAC,CAAgB,CAAA;QAEpB,OAAO;YACL,cAAc,gBAAA;YACd,gBAAgB,kBAAA;SACjB,CAAA;IACH,CAAC,CAAC,CAAA;AACJ,CAAC;AA9CD,0CA8CC"} {"version":3,"file":"predictionLayer.js","sourceRoot":"","sources":["../../src/faceDetectionNet/predictionLayer.ts"],"names":[],"mappings":";;AAAA,0CAA4C;AAE5C,2DAA0D;AAC1D,2DAA0D;AAG1D,yBACE,CAAc,EACd,MAAmB,EACnB,MAA6B;IAE7B,OAAO,EAAE,CAAC,IAAI,CAAC;QAEb,IAAM,KAAK,GAAG,uCAAkB,CAAC,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QAC1D,IAAM,KAAK,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,MAAM,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QAC9D,IAAM,KAAK,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,MAAM,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QAC9D,IAAM,KAAK,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,MAAM,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QAC9D,IAAM,KAAK,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,MAAM,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QAC9D,IAAM,KAAK,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,MAAM,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QAC9D,IAAM,KAAK,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,MAAM,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QAC9D,IAAM,KAAK,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,MAAM,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QAE9D,IAAM,cAAc,GAAG,uCAAkB,CAAC,MAAM,EAAE,MAAM,CAAC,eAAe,CAAC,CAAA;QACzE,IAAM,cAAc,GAAG,uCAAkB,CAAC,CAAC,EAAE,MAAM,CAAC,eAAe,CAAC,CAAA;QACpE,IAAM,cAAc,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,eAAe,CAAC,CAAA;QACxE,IAAM,cAAc,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,eAAe,CAAC,CAAA;QACxE,IAAM,cAAc,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,eAAe,CAAC,CAAA;QACxE,IAAM,cAAc,GAAG,uCAAkB,CAAC,KAAK,EAAE,MAAM,CAAC,eAAe,CAAC,CAAA;QAExE,IAAM,cAAc,GAAG,EAAE,CAAC,MAAM,CAAC;YAC/B,cAAc,CAAC,qBAAqB;YACpC,cAAc,CAAC,qBAAqB;YACpC,cAAc,CAAC,qBAAqB;YACpC,cAAc,CAAC,qBAAqB;YACpC,cAAc,CAAC,qBAAqB;YACpC,cAAc,CAAC,qBAAqB;SACrC,EAAE,CAAC,CAAgB,CAAA;QAEpB,IAAM,gBAAgB,GAAG,EAAE,CAAC,MAAM,CAAC;YACjC,cAAc,CAAC,eAAe;YAC9B,cAAc,CAAC,eAAe;YAC9B,cAAc,CAAC,eAAe;YAC9B,cAAc,CAAC,eAAe;YAC9B,cAAc,CAAC,eAAe;YAC9B,cAAc,CAAC,eAAe;SAC/B,EAAE,CAAC,CAAgB,CAAA;QAEpB,OAAO;YACL,cAAc,gBAAA;YACd,gBAAgB,kBAAA;SACjB,CAAA;IACH,CAAC,CAAC,CAAA;AACJ,CAAC;AA9CD,0CA8CC"}
\ No newline at end of file \ No newline at end of file
build/faceDetectionNet/types.d.ts
...@@ -13,39 +13,51 @@ export declare namespace MobileNetV1 { ...@@ -13,39 +13,51 @@ export declare namespace MobileNetV1 {
batch_norm_variance: tf.Tensor1D; batch_norm_variance: tf.Tensor1D;
}; };
type ConvPairParams = { type ConvPairParams = {
depthwise_conv_params: DepthwiseConvParams; depthwise_conv: DepthwiseConvParams;
pointwise_conv_params: PointwiseConvParams; pointwise_conv: PointwiseConvParams;
}; };
type Params = { type Params = {
conv_0_params: PointwiseConvParams; conv_0: PointwiseConvParams;
conv_pair_params: ConvPairParams[]; conv_1: ConvPairParams;
conv_2: ConvPairParams;
conv_3: ConvPairParams;
conv_4: ConvPairParams;
conv_5: ConvPairParams;
conv_6: ConvPairParams;
conv_7: ConvPairParams;
conv_8: ConvPairParams;
conv_9: ConvPairParams;
conv_10: ConvPairParams;
conv_11: ConvPairParams;
conv_12: ConvPairParams;
conv_13: ConvPairParams;
}; };
} }
export declare type BoxPredictionParams = { export declare type BoxPredictionParams = {
box_encoding_predictor_params: ConvParams; box_encoding_predictor: ConvParams;
class_predictor_params: ConvParams; class_predictor: ConvParams;
}; };
export declare type PredictionLayerParams = { export declare type PredictionLayerParams = {
conv_0_params: PointwiseConvParams; conv_0: PointwiseConvParams;
conv_1_params: PointwiseConvParams; conv_1: PointwiseConvParams;
conv_2_params: PointwiseConvParams; conv_2: PointwiseConvParams;
conv_3_params: PointwiseConvParams; conv_3: PointwiseConvParams;
conv_4_params: PointwiseConvParams; conv_4: PointwiseConvParams;
conv_5_params: PointwiseConvParams; conv_5: PointwiseConvParams;
conv_6_params: PointwiseConvParams; conv_6: PointwiseConvParams;
conv_7_params: PointwiseConvParams; conv_7: PointwiseConvParams;
box_predictor_0_params: BoxPredictionParams; box_predictor_0: BoxPredictionParams;
box_predictor_1_params: BoxPredictionParams; box_predictor_1: BoxPredictionParams;
box_predictor_2_params: BoxPredictionParams; box_predictor_2: BoxPredictionParams;
box_predictor_3_params: BoxPredictionParams; box_predictor_3: BoxPredictionParams;
box_predictor_4_params: BoxPredictionParams; box_predictor_4: BoxPredictionParams;
box_predictor_5_params: BoxPredictionParams; box_predictor_5: BoxPredictionParams;
}; };
export declare type OutputLayerParams = { export declare type OutputLayerParams = {
extra_dim: tf.Tensor3D; extra_dim: tf.Tensor3D;
}; };
export declare type NetParams = { export declare type NetParams = {
mobilenetv1_params: MobileNetV1.Params; mobilenetv1: MobileNetV1.Params;
prediction_layer_params: PredictionLayerParams; prediction_layer: PredictionLayerParams;
output_layer_params: OutputLayerParams; output_layer: OutputLayerParams;
}; };
build/faceLandmarkNet/FaceLandmarkNet.d.ts
...@@ -5,9 +5,22 @@ import { TNetInput } from '../types'; ...@@ -5,9 +5,22 @@ import { TNetInput } from '../types';
import { FaceLandmarks } from './FaceLandmarks'; import { FaceLandmarks } from './FaceLandmarks';
import { NetParams } from './types'; import { NetParams } from './types';
export declare class FaceLandmarkNet extends NeuralNetwork<NetParams> { export declare class FaceLandmarkNet extends NeuralNetwork<NetParams> {
load(weightsOrUrl: Float32Array | string | undefined): Promise<void>; constructor();
extractWeights(weights: Float32Array): void;
forwardInput(input: NetInput): tf.Tensor2D; forwardInput(input: NetInput): tf.Tensor2D;
forward(input: TNetInput): Promise<tf.Tensor2D>; forward(input: TNetInput): Promise<tf.Tensor2D>;
detectLandmarks(input: TNetInput): Promise<FaceLandmarks | FaceLandmarks[]>; detectLandmarks(input: TNetInput): Promise<FaceLandmarks | FaceLandmarks[]>;
protected loadQuantizedParams(uri: string | undefined): Promise<{
params: NetParams;
paramMappings: {
originalPath?: string | undefined;
paramPath: string;
}[];
}>;
protected extractParams(weights: Float32Array): {
params: NetParams;
paramMappings: {
originalPath?: string | undefined;
paramPath: string;
}[];
};
} }
build/faceLandmarkNet/FaceLandmarkNet.js
...@@ -21,57 +21,29 @@ function maxPool(x, strides) { ...@@ -21,57 +21,29 @@ function maxPool(x, strides) {
var FaceLandmarkNet = /** @class */ (function (_super) { var FaceLandmarkNet = /** @class */ (function (_super) {
tslib_1.__extends(FaceLandmarkNet, _super); tslib_1.__extends(FaceLandmarkNet, _super);
function FaceLandmarkNet() { function FaceLandmarkNet() {
return _super !== null && _super.apply(this, arguments) || this; return _super.call(this, 'FaceLandmarkNet') || this;
} }
FaceLandmarkNet.prototype.load = function (weightsOrUrl) {
return tslib_1.__awaiter(this, void 0, void 0, function () {
var _a, paramMappings, params;
return tslib_1.__generator(this, function (_b) {
switch (_b.label) {
case 0:
if (weightsOrUrl instanceof Float32Array) {
this.extractWeights(weightsOrUrl);
return [2 /*return*/];
}
if (weightsOrUrl && typeof weightsOrUrl !== 'string') {
throw new Error('FaceLandmarkNet.load - expected model uri, or weights as Float32Array');
}
return [4 /*yield*/, loadQuantizedParams_1.loadQuantizedParams(weightsOrUrl)];
case 1:
_a = _b.sent(), paramMappings = _a.paramMappings, params = _a.params;
this._paramMappings = paramMappings;
this._params = params;
return [2 /*return*/];
}
});
});
};
FaceLandmarkNet.prototype.extractWeights = function (weights) {
var _a = extractParams_1.extractParams(weights), paramMappings = _a.paramMappings, params = _a.params;
this._paramMappings = paramMappings;
this._params = params;
};
FaceLandmarkNet.prototype.forwardInput = function (input) { FaceLandmarkNet.prototype.forwardInput = function (input) {
var params = this._params; var params = this.params;
if (!params) { if (!params) {
throw new Error('FaceLandmarkNet - load model before inference'); throw new Error('FaceLandmarkNet - load model before inference');
} }
return tf.tidy(function () { return tf.tidy(function () {
var batchTensor = input.toBatchTensor(128, true); var batchTensor = input.toBatchTensor(128, true);
var out = conv(batchTensor, params.conv0_params); var out = conv(batchTensor, params.conv0);
out = maxPool(out); out = maxPool(out);
out = conv(out, params.conv1_params); out = conv(out, params.conv1);
out = conv(out, params.conv2_params); out = conv(out, params.conv2);
out = maxPool(out); out = maxPool(out);
out = conv(out, params.conv3_params); out = conv(out, params.conv3);
out = conv(out, params.conv4_params); out = conv(out, params.conv4);
out = maxPool(out); out = maxPool(out);
out = conv(out, params.conv5_params); out = conv(out, params.conv5);
out = conv(out, params.conv6_params); out = conv(out, params.conv6);
out = maxPool(out, [1, 1]); out = maxPool(out, [1, 1]);
out = conv(out, params.conv7_params); out = conv(out, params.conv7);
var fc0 = tf.relu(fullyConnectedLayer_1.fullyConnectedLayer(out.as2D(out.shape[0], -1), params.fc0_params)); var fc0 = tf.relu(fullyConnectedLayer_1.fullyConnectedLayer(out.as2D(out.shape[0], -1), params.fc0));
var fc1 = fullyConnectedLayer_1.fullyConnectedLayer(fc0, params.fc1_params); var fc1 = fullyConnectedLayer_1.fullyConnectedLayer(fc0, params.fc1);
var createInterleavedTensor = function (fillX, fillY) { var createInterleavedTensor = function (fillX, fillY) {
return tf.stack([ return tf.stack([
tf.fill([68], fillX), tf.fill([68], fillX),
...@@ -146,6 +118,12 @@ var FaceLandmarkNet = /** @class */ (function (_super) { ...@@ -146,6 +118,12 @@ var FaceLandmarkNet = /** @class */ (function (_super) {
}); });
}); });
}; };
FaceLandmarkNet.prototype.loadQuantizedParams = function (uri) {
return loadQuantizedParams_1.loadQuantizedParams(uri);
};
FaceLandmarkNet.prototype.extractParams = function (weights) {
return extractParams_1.extractParams(weights);
};
return FaceLandmarkNet; return FaceLandmarkNet;
}(NeuralNetwork_1.NeuralNetwork)); }(NeuralNetwork_1.NeuralNetwork));
exports.FaceLandmarkNet = FaceLandmarkNet; exports.FaceLandmarkNet = FaceLandmarkNet;
......
build/faceLandmarkNet/FaceLandmarkNet.js.map
{"version":3,"file":"FaceLandmarkNet.js","sourceRoot":"","sources":["../../src/faceLandmarkNet/FaceLandmarkNet.ts"],"names":[],"mappings":";;;AAAA,0CAA4C;AAE5C,kDAAiD;AACjD,0DAAyD;AAGzD,kCAAiC;AACjC,4CAA2C;AAE3C,kCAAkC;AAClC,iDAAgD;AAChD,iDAAgD;AAChD,6DAA4D;AAC5D,6DAA4D;AAG5D,cAAc,CAAc,EAAE,MAAkB;IAC9C,OAAO,qBAAS,CAAC,CAAC,EAAE,MAAM,EAAE,OAAO,EAAE,IAAI,CAAC,CAAA;AAC5C,CAAC;AAED,iBAAiB,CAAc,EAAE,OAAkC;IAAlC,wBAAA,EAAA,WAA6B,CAAC,EAAE,CAAC,CAAC;IACjE,OAAO,EAAE,CAAC,OAAO,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,OAAO,EAAE,OAAO,CAAC,CAAA;AAChD,CAAC;AAED;IAAqC,2CAAwB;IAA7D;;IA2HA,CAAC;IAzHc,8BAAI,GAAjB,UAAkB,YAA+C;;;;;;wBAC/D,IAAI,YAAY,YAAY,YAAY,EAAE;4BACxC,IAAI,CAAC,cAAc,CAAC,YAAY,CAAC,CAAA;4BACjC,sBAAM;yBACP;wBAED,IAAI,YAAY,IAAI,OAAO,YAAY,KAAK,QAAQ,EAAE;4BACpD,MAAM,IAAI,KAAK,CAAC,uEAAuE,CAAC,CAAA;yBACzF;wBAIG,qBAAM,yCAAmB,CAAC,YAAY,CAAC,EAAA;;wBAHrC,KAGF,SAAuC,EAFzC,aAAa,mBAAA,EACb,MAAM,YAAA;wBAGR,IAAI,CAAC,cAAc,GAAG,aAAa,CAAA;wBACnC,IAAI,CAAC,OAAO,GAAG,MAAM,CAAA;;;;;KACtB;IAEM,wCAAc,GAArB,UAAsB,OAAqB;QACnC,IAAA,2CAGoB,EAFxB,gCAAa,EACb,kBAAM,CACkB;QAE1B,IAAI,CAAC,cAAc,GAAG,aAAa,CAAA;QACnC,IAAI,CAAC,OAAO,GAAG,MAAM,CAAA;IACvB,CAAC;IAEM,sCAAY,GAAnB,UAAoB,KAAe;QACjC,IAAM,MAAM,GAAG,IAAI,CAAC,OAAO,CAAA;QAE3B,IAAI,CAAC,MAAM,EAAE;YACX,MAAM,IAAI,KAAK,CAAC,+CAA+C,CAAC,CAAA;SACjE;QAED,OAAO,EAAE,CAAC,IAAI,CAAC;YACb,IAAM,WAAW,GAAG,KAAK,CAAC,aAAa,CAAC,GAAG,EAAE,IAAI,CAAC,CAAA;YAElD,IAAI,GAAG,GAAG,IAAI,CAAC,WAAW,EAAE,MAAM,CAAC,YAAY,CAAC,CAAA;YAChD,GAAG,GAAG,OAAO,CAAC,GAAG,CAAC,CAAA;YAClB,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,MAAM,CAAC,YAAY,CAAC,CAAA;YACpC,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,MAAM,CAAC,YAAY,CAAC,CAAA;YACpC,GAAG,GAAG,OAAO,CAAC,GAAG,CAAC,CAAA;YAClB,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,MAAM,CAAC,YAAY,CAAC,CAAA;YACpC,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,MAAM,CAAC,YAAY,CAAC,CAAA;YACpC,GAAG,GAAG,OAAO,CAAC,GAAG,CAAC,CAAA;YAClB,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,MAAM,CAAC,YAAY,CAAC,CAAA;YACpC,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,MAAM,CAAC,YAAY,CAAC,CAAA;YACpC,GAAG,GAAG,OAAO,CAAC,GAAG,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;YAC1B,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,MAAM,CAAC,YAAY,CAAC,CAAA;YACpC,IAAM,GAAG,GAAG,EAAE,CAAC,IAAI,CAAC,yCAAmB,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,MAAM,CAAC,UAAU,CAAC,CAAC,CAAA;YACvF,IAAM,GAAG,GAAG,yCAAmB,CAAC,GAAG,EAAE,MAAM,CAAC,UAAU,CAAC,CAAA;YAEvD,IAAM,uBAAuB,GAAG,UAAC,KAAa,EAAE,KAAa;gBAC3D,OAAA,EAAE,CAAC,KAAK,CAAC;oBACP,EAAE,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC,EAAE,KAAK,CAAC;oBACpB,EAAE,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC,EAAE,KAAK,CAAC;iBACrB,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE;YAHzB,CAGyB,CAAA;YAE3B;;;cAGE;YAEF,IAAM,eAAe,GAAG,GAAG;iBACxB,GAAG,CAAC,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,SAAS,CAAC,EAAE,UAAC,CAAC,EAAE,QAAQ;gBAC3D,OAAA,uBAAuB,CACrB,KAAK,CAAC,WAAW,CAAC,QAAQ,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,aAAa,CAAC,QAAQ,CAAC,EAC7D,KAAK,CAAC,WAAW,CAAC,QAAQ,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,cAAc,CAAC,QAAQ,CAAC,CAC/D;YAHD,CAGC,CACF,CAAC,CAAC;iBACF,GAAG,CAAC,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,SAAS,CAAC,EAAE,UAAC,CAAC,EAAE,QAAQ;gBAC3D,OAAA,uBAAuB,CACrB,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,WAAW,CAAC,QAAQ,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,EAC7C,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,WAAW,CAAC,QAAQ,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAC9C;YAHD,CAGC,CACF,CAAC,CAAC;iBACF,GAAG,CAAC,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,SAAS,CAAC,EAAE,UAAC,CAAC,EAAE,QAAQ;gBAC3D,OAAA,uBAAuB,CACrB,KAAK,CAAC,aAAa,CAAC,QAAQ,CAAC,EAC7B,KAAK,CAAC,cAAc,CAAC,QAAQ,CAAC,CAC/B;YAHD,CAGC,CACF,CAAC,CAAC,CAAA;YAEL,OAAO,eAA8B,CAAA;QACvC,CAAC,CAAC,CAAA;IACJ,CAAC;IAEY,iCAAO,GAApB,UAAqB,KAAgB;;;;;;wBAC5B,KAAA,IAAI,CAAC,YAAY,CAAA;wBAAC,qBAAM,uBAAU,CAAC,KAAK,EAAE,IAAI,CAAC,EAAA;4BAAtD,sBAAO,SAAA,IAAI,GAAc,SAA6B,EAAC,EAAA;;;;KACxD;IAEY,yCAAe,GAA5B,UAA6B,KAAgB;;;;;;4BAC1B,qBAAM,uBAAU,CAAC,KAAK,EAAE,IAAI,CAAC,EAAA;;wBAAxC,QAAQ,GAAG,SAA6B;wBAExC,eAAe,GAAG,EAAE,CAAC,IAAI,CAC7B,cAAM,OAAA,EAAE,CAAC,OAAO,CAAC,KAAI,CAAC,YAAY,CAAC,QAAQ,CAAC,CAAC,EAAvC,CAAuC,CAC9C,CAAA;wBAEyB,qBAAM,OAAO,CAAC,GAAG,CAAC,eAAe,CAAC,GAAG,CAC7D,UAAO,cAAc,EAAE,QAAQ;;;;;4CACN,KAAA,CAAA,KAAA,KAAK,CAAA,CAAC,IAAI,CAAA;4CAAC,qBAAM,cAAc,CAAC,IAAI,EAAE,EAAA;;4CAAvD,cAAc,GAAG,cAAW,SAA2B,EAAC;4CACxD,OAAO,GAAG,cAAc,CAAC,MAAM,CAAC,UAAC,CAAC,EAAE,CAAC,IAAK,OAAA,cAAM,CAAC,CAAC,CAAC,EAAT,CAAS,CAAC,CAAA;4CACpD,OAAO,GAAG,cAAc,CAAC,MAAM,CAAC,UAAC,CAAC,EAAE,CAAC,IAAK,OAAA,CAAC,cAAM,CAAC,CAAC,CAAC,EAAV,CAAU,CAAC,CAAA;4CAE3D,sBAAO,IAAI,6BAAa,CACtB,KAAK,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,UAAC,CAAC,EAAE,CAAC,IAAK,OAAA,IAAI,aAAK,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,OAAO,CAAC,CAAC,CAAC,CAAC,EAAjC,CAAiC,CAAC,EAClE;oDACE,MAAM,EAAE,QAAQ,CAAC,cAAc,CAAC,QAAQ,CAAC;oDACzC,KAAK,EAAG,QAAQ,CAAC,aAAa,CAAC,QAAQ,CAAC;iDACzC,CACF,EAAA;;;iCACF,CACF,CAAC,EAAA;;wBAdI,iBAAiB,GAAG,SAcxB;wBAEF,eAAe,CAAC,OAAO,CAAC,UAAA,CAAC,IAAI,OAAA,CAAC,CAAC,OAAO,EAAE,EAAX,CAAW,CAAC,CAAA;wBAEzC,sBAAO,QAAQ,CAAC,YAAY;gCAC1B,CAAC,CAAC,iBAAiB;gCACnB,CAAC,CAAC,iBAAiB,CAAC,CAAC,CAAC,EAAA;;;;KACzB;IACH,sBAAC;AAAD,CAAC,AA3HD,CAAqC,6BAAa,GA2HjD;AA3HY,0CAAe"} {"version":3,"file":"FaceLandmarkNet.js","sourceRoot":"","sources":["../../src/faceLandmarkNet/FaceLandmarkNet.ts"],"names":[],"mappings":";;;AAAA,0CAA4C;AAE5C,kDAAiD;AACjD,0DAAyD;AAGzD,kCAAiC;AACjC,4CAA2C;AAE3C,kCAAkC;AAClC,iDAAgD;AAChD,iDAAgD;AAChD,6DAA4D;AAC5D,6DAA4D;AAG5D,cAAc,CAAc,EAAE,MAAkB;IAC9C,OAAO,qBAAS,CAAC,CAAC,EAAE,MAAM,EAAE,OAAO,EAAE,IAAI,CAAC,CAAA;AAC5C,CAAC;AAED,iBAAiB,CAAc,EAAE,OAAkC;IAAlC,wBAAA,EAAA,WAA6B,CAAC,EAAE,CAAC,CAAC;IACjE,OAAO,EAAE,CAAC,OAAO,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,OAAO,EAAE,OAAO,CAAC,CAAA;AAChD,CAAC;AAED;IAAqC,2CAAwB;IAE3D;eACE,kBAAM,iBAAiB,CAAC;IAC1B,CAAC;IAEM,sCAAY,GAAnB,UAAoB,KAAe;QAEzB,IAAA,oBAAM,CAAS;QAEvB,IAAI,CAAC,MAAM,EAAE;YACX,MAAM,IAAI,KAAK,CAAC,+CAA+C,CAAC,CAAA;SACjE;QAED,OAAO,EAAE,CAAC,IAAI,CAAC;YACb,IAAM,WAAW,GAAG,KAAK,CAAC,aAAa,CAAC,GAAG,EAAE,IAAI,CAAC,CAAA;YAElD,IAAI,GAAG,GAAG,IAAI,CAAC,WAAW,EAAE,MAAM,CAAC,KAAK,CAAC,CAAA;YACzC,GAAG,GAAG,OAAO,CAAC,GAAG,CAAC,CAAA;YAClB,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,MAAM,CAAC,KAAK,CAAC,CAAA;YAC7B,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,MAAM,CAAC,KAAK,CAAC,CAAA;YAC7B,GAAG,GAAG,OAAO,CAAC,GAAG,CAAC,CAAA;YAClB,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,MAAM,CAAC,KAAK,CAAC,CAAA;YAC7B,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,MAAM,CAAC,KAAK,CAAC,CAAA;YAC7B,GAAG,GAAG,OAAO,CAAC,GAAG,CAAC,CAAA;YAClB,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,MAAM,CAAC,KAAK,CAAC,CAAA;YAC7B,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,MAAM,CAAC,KAAK,CAAC,CAAA;YAC7B,GAAG,GAAG,OAAO,CAAC,GAAG,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;YAC1B,GAAG,GAAG,IAAI,CAAC,GAAG,EAAE,MAAM,CAAC,KAAK,CAAC,CAAA;YAC7B,IAAM,GAAG,GAAG,EAAE,CAAC,IAAI,CAAC,yCAAmB,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,MAAM,CAAC,GAAG,CAAC,CAAC,CAAA;YAChF,IAAM,GAAG,GAAG,yCAAmB,CAAC,GAAG,EAAE,MAAM,CAAC,GAAG,CAAC,CAAA;YAEhD,IAAM,uBAAuB,GAAG,UAAC,KAAa,EAAE,KAAa;gBAC3D,OAAA,EAAE,CAAC,KAAK,CAAC;oBACP,EAAE,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC,EAAE,KAAK,CAAC;oBACpB,EAAE,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC,EAAE,KAAK,CAAC;iBACrB,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE;YAHzB,CAGyB,CAAA;YAE3B;;;cAGE;YAEF,IAAM,eAAe,GAAG,GAAG;iBACxB,GAAG,CAAC,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,SAAS,CAAC,EAAE,UAAC,CAAC,EAAE,QAAQ;gBAC3D,OAAA,uBAAuB,CACrB,KAAK,CAAC,WAAW,CAAC,QAAQ,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,aAAa,CAAC,QAAQ,CAAC,EAC7D,KAAK,CAAC,WAAW,CAAC,QAAQ,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,cAAc,CAAC,QAAQ,CAAC,CAC/D;YAHD,CAGC,CACF,CAAC,CAAC;iBACF,GAAG,CAAC,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,SAAS,CAAC,EAAE,UAAC,CAAC,EAAE,QAAQ;gBAC3D,OAAA,uBAAuB,CACrB,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,WAAW,CAAC,QAAQ,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,EAC7C,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,WAAW,CAAC,QAAQ,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAC9C;YAHD,CAGC,CACF,CAAC,CAAC;iBACF,GAAG,CAAC,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,SAAS,CAAC,EAAE,UAAC,CAAC,EAAE,QAAQ;gBAC3D,OAAA,uBAAuB,CACrB,KAAK,CAAC,aAAa,CAAC,QAAQ,CAAC,EAC7B,KAAK,CAAC,cAAc,CAAC,QAAQ,CAAC,CAC/B;YAHD,CAGC,CACF,CAAC,CAAC,CAAA;YAEL,OAAO,eAA8B,CAAA;QACvC,CAAC,CAAC,CAAA;IACJ,CAAC;IAEY,iCAAO,GAApB,UAAqB,KAAgB;;;;;;wBAC5B,KAAA,IAAI,CAAC,YAAY,CAAA;wBAAC,qBAAM,uBAAU,CAAC,KAAK,EAAE,IAAI,CAAC,EAAA;4BAAtD,sBAAO,SAAA,IAAI,GAAc,SAA6B,EAAC,EAAA;;;;KACxD;IAEY,yCAAe,GAA5B,UAA6B,KAAgB;;;;;;4BAC1B,qBAAM,uBAAU,CAAC,KAAK,EAAE,IAAI,CAAC,EAAA;;wBAAxC,QAAQ,GAAG,SAA6B;wBAExC,eAAe,GAAG,EAAE,CAAC,IAAI,CAC7B,cAAM,OAAA,EAAE,CAAC,OAAO,CAAC,KAAI,CAAC,YAAY,CAAC,QAAQ,CAAC,CAAC,EAAvC,CAAuC,CAC9C,CAAA;wBAEyB,qBAAM,OAAO,CAAC,GAAG,CAAC,eAAe,CAAC,GAAG,CAC7D,UAAO,cAAc,EAAE,QAAQ;;;;;4CACN,KAAA,CAAA,KAAA,KAAK,CAAA,CAAC,IAAI,CAAA;4CAAC,qBAAM,cAAc,CAAC,IAAI,EAAE,EAAA;;4CAAvD,cAAc,GAAG,cAAW,SAA2B,EAAC;4CACxD,OAAO,GAAG,cAAc,CAAC,MAAM,CAAC,UAAC,CAAC,EAAE,CAAC,IAAK,OAAA,cAAM,CAAC,CAAC,CAAC,EAAT,CAAS,CAAC,CAAA;4CACpD,OAAO,GAAG,cAAc,CAAC,MAAM,CAAC,UAAC,CAAC,EAAE,CAAC,IAAK,OAAA,CAAC,cAAM,CAAC,CAAC,CAAC,EAAV,CAAU,CAAC,CAAA;4CAE3D,sBAAO,IAAI,6BAAa,CACtB,KAAK,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,UAAC,CAAC,EAAE,CAAC,IAAK,OAAA,IAAI,aAAK,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,OAAO,CAAC,CAAC,CAAC,CAAC,EAAjC,CAAiC,CAAC,EAClE;oDACE,MAAM,EAAE,QAAQ,CAAC,cAAc,CAAC,QAAQ,CAAC;oDACzC,KAAK,EAAG,QAAQ,CAAC,aAAa,CAAC,QAAQ,CAAC;iDACzC,CACF,EAAA;;;iCACF,CACF,CAAC,EAAA;;wBAdI,iBAAiB,GAAG,SAcxB;wBAEF,eAAe,CAAC,OAAO,CAAC,UAAA,CAAC,IAAI,OAAA,CAAC,CAAC,OAAO,EAAE,EAAX,CAAW,CAAC,CAAA;wBAEzC,sBAAO,QAAQ,CAAC,YAAY;gCAC1B,CAAC,CAAC,iBAAiB;gCACnB,CAAC,CAAC,iBAAiB,CAAC,CAAC,CAAC,EAAA;;;;KACzB;IAES,6CAAmB,GAA7B,UAA8B,GAAuB;QACnD,OAAO,yCAAmB,CAAC,GAAG,CAAC,CAAA;IACjC,CAAC;IAES,uCAAa,GAAvB,UAAwB,OAAqB;QAC3C,OAAO,6BAAa,CAAC,OAAO,CAAC,CAAA;IAC/B,CAAC;IACH,sBAAC;AAAD,CAAC,AA5GD,CAAqC,6BAAa,GA4GjD;AA5GY,0CAAe"}
\ No newline at end of file \ No newline at end of file
build/faceLandmarkNet/extractParams.js
"use strict"; "use strict";
Object.defineProperty(exports, "__esModule", { value: true }); Object.defineProperty(exports, "__esModule", { value: true });
var tf = require("@tensorflow/tfjs-core"); var tf = require("@tensorflow/tfjs-core");
var extractConvParamsFactory_1 = require("../commons/extractConvParamsFactory");
var extractWeightsFactory_1 = require("../commons/extractWeightsFactory"); var extractWeightsFactory_1 = require("../commons/extractWeightsFactory");
function extractParams(weights) { function extractParams(weights) {
var paramMappings = []; var paramMappings = [];
var _a = extractWeightsFactory_1.extractWeightsFactory(weights), extractWeights = _a.extractWeights, getRemainingWeights = _a.getRemainingWeights; var _a = extractWeightsFactory_1.extractWeightsFactory(weights), extractWeights = _a.extractWeights, getRemainingWeights = _a.getRemainingWeights;
var extractConvParams = extractConvParamsFactory_1.extractConvParamsFactory(extractWeights, paramMappings); function extractConvParams(channelsIn, channelsOut, filterSize, mappedPrefix) {
var filters = tf.tensor4d(extractWeights(channelsIn * channelsOut * filterSize * filterSize), [filterSize, filterSize, channelsIn, channelsOut]);
var bias = tf.tensor1d(extractWeights(channelsOut));
paramMappings.push({ paramPath: mappedPrefix + "/filters" }, { paramPath: mappedPrefix + "/bias" });
return { filters: filters, bias: bias };
}
function extractFcParams(channelsIn, channelsOut, mappedPrefix) { function extractFcParams(channelsIn, channelsOut, mappedPrefix) {
var fc_weights = tf.tensor2d(extractWeights(channelsIn * channelsOut), [channelsIn, channelsOut]); var fc_weights = tf.tensor2d(extractWeights(channelsIn * channelsOut), [channelsIn, channelsOut]);
var fc_bias = tf.tensor1d(extractWeights(channelsOut)); var fc_bias = tf.tensor1d(extractWeights(channelsOut));
...@@ -16,32 +20,32 @@ function extractParams(weights) { ...@@ -16,32 +20,32 @@ function extractParams(weights) {
bias: fc_bias bias: fc_bias
}; };
} }
var conv0_params = extractConvParams(3, 32, 3, 'conv0_params'); var conv0 = extractConvParams(3, 32, 3, 'conv0');
var conv1_params = extractConvParams(32, 64, 3, 'conv1_params'); var conv1 = extractConvParams(32, 64, 3, 'conv1');
var conv2_params = extractConvParams(64, 64, 3, 'conv2_params'); var conv2 = extractConvParams(64, 64, 3, 'conv2');
var conv3_params = extractConvParams(64, 64, 3, 'conv3_params'); var conv3 = extractConvParams(64, 64, 3, 'conv3');
var conv4_params = extractConvParams(64, 64, 3, 'conv4_params'); var conv4 = extractConvParams(64, 64, 3, 'conv4');
var conv5_params = extractConvParams(64, 128, 3, 'conv5_params'); var conv5 = extractConvParams(64, 128, 3, 'conv5');
var conv6_params = extractConvParams(128, 128, 3, 'conv6_params'); var conv6 = extractConvParams(128, 128, 3, 'conv6');
var conv7_params = extractConvParams(128, 256, 3, 'conv7_params'); var conv7 = extractConvParams(128, 256, 3, 'conv7');
var fc0_params = extractFcParams(6400, 1024, 'fc0_params'); var fc0 = extractFcParams(6400, 1024, 'fc0');
var fc1_params = extractFcParams(1024, 136, 'fc1_params'); var fc1 = extractFcParams(1024, 136, 'fc1');
if (getRemainingWeights().length !== 0) { if (getRemainingWeights().length !== 0) {
throw new Error("weights remaing after extract: " + getRemainingWeights().length); throw new Error("weights remaing after extract: " + getRemainingWeights().length);
} }
return { return {
paramMappings: paramMappings, paramMappings: paramMappings,
params: { params: {
conv0_params: conv0_params, conv0: conv0,
conv1_params: conv1_params, conv1: conv1,
conv2_params: conv2_params, conv2: conv2,
conv3_params: conv3_params, conv3: conv3,
conv4_params: conv4_params, conv4: conv4,
conv5_params: conv5_params, conv5: conv5,
conv6_params: conv6_params, conv6: conv6,
conv7_params: conv7_params, conv7: conv7,
fc0_params: fc0_params, fc0: fc0,
fc1_params: fc1_params fc1: fc1
} }
}; };
} }
......
build/faceLandmarkNet/extractParams.js.map
{"version":3,"file":"extractParams.js","sourceRoot":"","sources":["../../src/faceLandmarkNet/extractParams.ts"],"names":[],"mappings":";;AAAA,0CAA4C;AAE5C,gFAA+E;AAC/E,0EAAyE;AAIzE,uBAA8B,OAAqB;IACjD,IAAM,aAAa,GAAmB,EAAE,CAAA;IAElC,IAAA,2DAG4B,EAFhC,kCAAc,EACd,4CAAmB,CACa;IAElC,IAAM,iBAAiB,GAAG,mDAAwB,CAAC,cAAc,EAAE,aAAa,CAAC,CAAA;IAEjF,yBAAyB,UAAkB,EAAE,WAAmB,EAAE,YAAoB;QACpF,IAAM,UAAU,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,UAAU,GAAG,WAAW,CAAC,EAAE,CAAC,UAAU,EAAE,WAAW,CAAC,CAAC,CAAA;QACnG,IAAM,OAAO,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,WAAW,CAAC,CAAC,CAAA;QAExD,aAAa,CAAC,IAAI,CAChB,EAAE,SAAS,EAAK,YAAY,aAAU,EAAE,EACxC,EAAE,SAAS,EAAK,YAAY,UAAO,EAAE,CACtC,CAAA;QAED,OAAO;YACL,OAAO,EAAE,UAAU;YACnB,IAAI,EAAE,OAAO;SACd,CAAA;IACH,CAAC;IAED,IAAM,YAAY,GAAG,iBAAiB,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,cAAc,CAAC,CAAA;IAChE,IAAM,YAAY,GAAG,iBAAiB,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC,EAAE,cAAc,CAAC,CAAA;IACjE,IAAM,YAAY,GAAG,iBAAiB,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC,EAAE,cAAc,CAAC,CAAA;IACjE,IAAM,YAAY,GAAG,iBAAiB,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC,EAAE,cAAc,CAAC,CAAA;IACjE,IAAM,YAAY,GAAG,iBAAiB,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC,EAAE,cAAc,CAAC,CAAA;IACjE,IAAM,YAAY,GAAG,iBAAiB,CAAC,EAAE,EAAE,GAAG,EAAE,CAAC,EAAE,cAAc,CAAC,CAAA;IAClE,IAAM,YAAY,GAAG,iBAAiB,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC,EAAE,cAAc,CAAC,CAAA;IACnE,IAAM,YAAY,GAAG,iBAAiB,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC,EAAE,cAAc,CAAC,CAAA;IACnE,IAAM,UAAU,GAAG,eAAe,CAAC,IAAI,EAAE,IAAI,EAAE,YAAY,CAAC,CAAA;IAC5D,IAAM,UAAU,GAAG,eAAe,CAAC,IAAI,EAAE,GAAG,EAAE,YAAY,CAAC,CAAA;IAE3D,IAAI,mBAAmB,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE;QACtC,MAAM,IAAI,KAAK,CAAC,oCAAkC,mBAAmB,EAAE,CAAC,MAAQ,CAAC,CAAA;KAClF;IAED,OAAO;QACL,aAAa,eAAA;QACb,MAAM,EAAE;YACN,YAAY,cAAA;YACZ,YAAY,cAAA;YACZ,YAAY,cAAA;YACZ,YAAY,cAAA;YACZ,YAAY,cAAA;YACZ,YAAY,cAAA;YACZ,YAAY,cAAA;YACZ,YAAY,cAAA;YACZ,UAAU,YAAA;YACV,UAAU,YAAA;SACX;KACF,CAAA;AACH,CAAC;AAvDD,sCAuDC"} {"version":3,"file":"extractParams.js","sourceRoot":"","sources":["../../src/faceLandmarkNet/extractParams.ts"],"names":[],"mappings":";;AAAA,0CAA4C;AAE5C,0EAAyE;AAIzE,uBAA8B,OAAqB;IAEjD,IAAM,aAAa,GAAmB,EAAE,CAAA;IAElC,IAAA,2DAG4B,EAFhC,kCAAc,EACd,4CAAmB,CACa;IAElC,2BACE,UAAkB,EAClB,WAAmB,EACnB,UAAkB,EAClB,YAAoB;QAGpB,IAAM,OAAO,GAAG,EAAE,CAAC,QAAQ,CACzB,cAAc,CAAC,UAAU,GAAG,WAAW,GAAG,UAAU,GAAG,UAAU,CAAC,EAClE,CAAC,UAAU,EAAE,UAAU,EAAE,UAAU,EAAE,WAAW,CAAC,CAClD,CAAA;QACD,IAAM,IAAI,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,WAAW,CAAC,CAAC,CAAA;QAErD,aAAa,CAAC,IAAI,CAChB,EAAE,SAAS,EAAK,YAAY,aAAU,EAAE,EACxC,EAAE,SAAS,EAAK,YAAY,UAAO,EAAE,CACtC,CAAA;QAED,OAAO,EAAE,OAAO,SAAA,EAAE,IAAI,MAAA,EAAE,CAAA;IAC1B,CAAC;IAED,yBAAyB,UAAkB,EAAE,WAAmB,EAAE,YAAoB;QAEpF,IAAM,UAAU,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,UAAU,GAAG,WAAW,CAAC,EAAE,CAAC,UAAU,EAAE,WAAW,CAAC,CAAC,CAAA;QACnG,IAAM,OAAO,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,WAAW,CAAC,CAAC,CAAA;QAExD,aAAa,CAAC,IAAI,CAChB,EAAE,SAAS,EAAK,YAAY,aAAU,EAAE,EACxC,EAAE,SAAS,EAAK,YAAY,UAAO,EAAE,CACtC,CAAA;QAED,OAAO;YACL,OAAO,EAAE,UAAU;YACnB,IAAI,EAAE,OAAO;SACd,CAAA;IACH,CAAC;IAED,IAAM,KAAK,GAAG,iBAAiB,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,OAAO,CAAC,CAAA;IAClD,IAAM,KAAK,GAAG,iBAAiB,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC,EAAE,OAAO,CAAC,CAAA;IACnD,IAAM,KAAK,GAAG,iBAAiB,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC,EAAE,OAAO,CAAC,CAAA;IACnD,IAAM,KAAK,GAAG,iBAAiB,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC,EAAE,OAAO,CAAC,CAAA;IACnD,IAAM,KAAK,GAAG,iBAAiB,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC,EAAE,OAAO,CAAC,CAAA;IACnD,IAAM,KAAK,GAAG,iBAAiB,CAAC,EAAE,EAAE,GAAG,EAAE,CAAC,EAAE,OAAO,CAAC,CAAA;IACpD,IAAM,KAAK,GAAG,iBAAiB,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC,EAAE,OAAO,CAAC,CAAA;IACrD,IAAM,KAAK,GAAG,iBAAiB,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC,EAAE,OAAO,CAAC,CAAA;IACrD,IAAM,GAAG,GAAG,eAAe,CAAC,IAAI,EAAE,IAAI,EAAE,KAAK,CAAC,CAAA;IAC9C,IAAM,GAAG,GAAG,eAAe,CAAC,IAAI,EAAE,GAAG,EAAE,KAAK,CAAC,CAAA;IAE7C,IAAI,mBAAmB,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE;QACtC,MAAM,IAAI,KAAK,CAAC,oCAAkC,mBAAmB,EAAE,CAAC,MAAQ,CAAC,CAAA;KAClF;IAED,OAAO;QACL,aAAa,eAAA;QACb,MAAM,EAAE;YACN,KAAK,OAAA;YACL,KAAK,OAAA;YACL,KAAK,OAAA;YACL,KAAK,OAAA;YACL,KAAK,OAAA;YACL,KAAK,OAAA;YACL,KAAK,OAAA;YACL,KAAK,OAAA;YACL,GAAG,KAAA;YACH,GAAG,KAAA;SACJ;KACF,CAAA;AACH,CAAC;AA5ED,sCA4EC"}
\ No newline at end of file \ No newline at end of file
build/faceLandmarkNet/loadQuantizedParams.js
"use strict"; "use strict";
Object.defineProperty(exports, "__esModule", { value: true }); Object.defineProperty(exports, "__esModule", { value: true });
var tslib_1 = require("tslib"); var tslib_1 = require("tslib");
var extractWeightEntry_1 = require("../commons/extractWeightEntry"); var disposeUnusedWeightTensors_1 = require("../commons/disposeUnusedWeightTensors");
var extractWeightEntryFactory_1 = require("../commons/extractWeightEntryFactory");
var loadWeightMap_1 = require("../commons/loadWeightMap"); var loadWeightMap_1 = require("../commons/loadWeightMap");
var DEFAULT_MODEL_NAME = 'face_landmark_68_model'; var DEFAULT_MODEL_NAME = 'face_landmark_68_model';
function extractorsFactory(weightMap, paramMappings) { function extractorsFactory(weightMap, paramMappings) {
var extractWeightEntry = extractWeightEntryFactory_1.extractWeightEntryFactory(weightMap, paramMappings);
function extractConvParams(prefix, mappedPrefix) { function extractConvParams(prefix, mappedPrefix) {
var filtersEntry = extractWeightEntry_1.extractWeightEntry(weightMap, prefix + "/kernel", 4); var filters = extractWeightEntry(prefix + "/kernel", 4, mappedPrefix + "/filters");
var biasEntry = extractWeightEntry_1.extractWeightEntry(weightMap, prefix + "/bias", 1); var bias = extractWeightEntry(prefix + "/bias", 1, mappedPrefix + "/bias");
paramMappings.push({ originalPath: filtersEntry.path, paramPath: mappedPrefix + "/filters" }, { originalPath: biasEntry.path, paramPath: mappedPrefix + "/bias" }); return { filters: filters, bias: bias };
return {
filters: filtersEntry.tensor,
bias: biasEntry.tensor
};
} }
function extractFcParams(prefix, mappedPrefix) { function extractFcParams(prefix, mappedPrefix) {
var weightsEntry = extractWeightEntry_1.extractWeightEntry(weightMap, prefix + "/kernel", 2); var weights = extractWeightEntry(prefix + "/kernel", 2, mappedPrefix + "/weights");
var biasEntry = extractWeightEntry_1.extractWeightEntry(weightMap, prefix + "/bias", 1); var bias = extractWeightEntry(prefix + "/bias", 1, mappedPrefix + "/bias");
paramMappings.push({ originalPath: weightsEntry.path, paramPath: mappedPrefix + "/weights" }, { originalPath: biasEntry.path, paramPath: mappedPrefix + "/bias" }); return { weights: weights, bias: bias };
return {
weights: weightsEntry.tensor,
bias: biasEntry.tensor
};
} }
return { return {
extractConvParams: extractConvParams, extractConvParams: extractConvParams,
...@@ -39,17 +33,18 @@ function loadQuantizedParams(uri) { ...@@ -39,17 +33,18 @@ function loadQuantizedParams(uri) {
paramMappings = []; paramMappings = [];
_a = extractorsFactory(weightMap, paramMappings), extractConvParams = _a.extractConvParams, extractFcParams = _a.extractFcParams; _a = extractorsFactory(weightMap, paramMappings), extractConvParams = _a.extractConvParams, extractFcParams = _a.extractFcParams;
params = { params = {
conv0_params: extractConvParams('conv2d_0', 'conv0_params'), conv0: extractConvParams('conv2d_0', 'conv0'),
conv1_params: extractConvParams('conv2d_1', 'conv1_params'), conv1: extractConvParams('conv2d_1', 'conv1'),
conv2_params: extractConvParams('conv2d_2', 'conv2_params'), conv2: extractConvParams('conv2d_2', 'conv2'),
conv3_params: extractConvParams('conv2d_3', 'conv3_params'), conv3: extractConvParams('conv2d_3', 'conv3'),
conv4_params: extractConvParams('conv2d_4', 'conv4_params'), conv4: extractConvParams('conv2d_4', 'conv4'),
conv5_params: extractConvParams('conv2d_5', 'conv5_params'), conv5: extractConvParams('conv2d_5', 'conv5'),
conv6_params: extractConvParams('conv2d_6', 'conv6_params'), conv6: extractConvParams('conv2d_6', 'conv6'),
conv7_params: extractConvParams('conv2d_7', 'conv7_params'), conv7: extractConvParams('conv2d_7', 'conv7'),
fc0_params: extractFcParams('dense', 'fc0_params'), fc0: extractFcParams('dense', 'fc0'),
fc1_params: extractFcParams('logits', 'fc1_params') fc1: extractFcParams('logits', 'fc1')
}; };
disposeUnusedWeightTensors_1.disposeUnusedWeightTensors(weightMap, paramMappings);
return [2 /*return*/, { params: params, paramMappings: paramMappings }]; return [2 /*return*/, { params: params, paramMappings: paramMappings }];
} }
}); });
......
build/faceLandmarkNet/loadQuantizedParams.js.map
{"version":3,"file":"loadQuantizedParams.js","sourceRoot":"","sources":["../../src/faceLandmarkNet/loadQuantizedParams.ts"],"names":[],"mappings":";;;AAEA,oEAAmE;AACnE,0DAAyD;AAIzD,IAAM,kBAAkB,GAAG,wBAAwB,CAAA;AAEnD,2BAA2B,SAAc,EAAE,aAA6B;IAEtE,2BAA2B,MAAc,EAAE,YAAoB;QAC7D,IAAM,YAAY,GAAG,uCAAkB,CAAC,SAAS,EAAK,MAAM,YAAS,EAAE,CAAC,CAAC,CAAA;QACzE,IAAM,SAAS,GAAG,uCAAkB,CAAC,SAAS,EAAK,MAAM,UAAO,EAAE,CAAC,CAAC,CAAA;QACpE,aAAa,CAAC,IAAI,CAChB,EAAE,YAAY,EAAE,YAAY,CAAC,IAAI,EAAE,SAAS,EAAK,YAAY,aAAU,EAAE,EACzE,EAAE,YAAY,EAAE,SAAS,CAAC,IAAI,EAAE,SAAS,EAAK,YAAY,UAAO,EAAE,CACpE,CAAA;QACD,OAAO;YACL,OAAO,EAAE,YAAY,CAAC,MAAqB;YAC3C,IAAI,EAAE,SAAS,CAAC,MAAqB;SACtC,CAAA;IACH,CAAC;IAED,yBAAyB,MAAc,EAAE,YAAoB;QAC3D,IAAM,YAAY,GAAG,uCAAkB,CAAC,SAAS,EAAK,MAAM,YAAS,EAAE,CAAC,CAAC,CAAA;QACzE,IAAM,SAAS,GAAG,uCAAkB,CAAC,SAAS,EAAK,MAAM,UAAO,EAAE,CAAC,CAAC,CAAA;QACpE,aAAa,CAAC,IAAI,CAChB,EAAE,YAAY,EAAE,YAAY,CAAC,IAAI,EAAE,SAAS,EAAK,YAAY,aAAU,EAAE,EACzE,EAAE,YAAY,EAAE,SAAS,CAAC,IAAI,EAAE,SAAS,EAAK,YAAY,UAAO,EAAE,CACpE,CAAA;QACD,OAAO;YACL,OAAO,EAAE,YAAY,CAAC,MAAqB;YAC3C,IAAI,EAAE,SAAS,CAAC,MAAqB;SACtC,CAAA;IACH,CAAC;IAED,OAAO;QACL,iBAAiB,mBAAA;QACjB,eAAe,iBAAA;KAChB,CAAA;AACH,CAAC;AAED,6BACE,GAAuB;;;;;wBAGL,qBAAM,6BAAa,CAAC,GAAG,EAAE,kBAAkB,CAAC,EAAA;;oBAAxD,SAAS,GAAG,SAA4C;oBACxD,aAAa,GAAmB,EAAE,CAAA;oBAElC,KAGF,iBAAiB,CAAC,SAAS,EAAE,aAAa,CAAC,EAF7C,iBAAiB,uBAAA,EACjB,eAAe,qBAAA,CAC8B;oBAEzC,MAAM,GAAG;wBACb,YAAY,EAAE,iBAAiB,CAAC,UAAU,EAAE,cAAc,CAAC;wBAC3D,YAAY,EAAE,iBAAiB,CAAC,UAAU,EAAE,cAAc,CAAC;wBAC3D,YAAY,EAAE,iBAAiB,CAAC,UAAU,EAAE,cAAc,CAAC;wBAC3D,YAAY,EAAE,iBAAiB,CAAC,UAAU,EAAE,cAAc,CAAC;wBAC3D,YAAY,EAAE,iBAAiB,CAAC,UAAU,EAAE,cAAc,CAAC;wBAC3D,YAAY,EAAE,iBAAiB,CAAC,UAAU,EAAE,cAAc,CAAC;wBAC3D,YAAY,EAAE,iBAAiB,CAAC,UAAU,EAAE,cAAc,CAAC;wBAC3D,YAAY,EAAE,iBAAiB,CAAC,UAAU,EAAE,cAAc,CAAC;wBAC3D,UAAU,EAAE,eAAe,CAAC,OAAO,EAAE,YAAY,CAAC;wBAClD,UAAU,EAAE,eAAe,CAAC,QAAQ,EAAE,YAAY,CAAC;qBACpD,CAAA;oBAED,sBAAO,EAAE,MAAM,QAAA,EAAE,aAAa,eAAA,EAAE,EAAA;;;;CACjC;AA1BD,kDA0BC"} {"version":3,"file":"loadQuantizedParams.js","sourceRoot":"","sources":["../../src/faceLandmarkNet/loadQuantizedParams.ts"],"names":[],"mappings":";;;AAEA,oFAAmF;AACnF,kFAAiF;AACjF,0DAAyD;AAIzD,IAAM,kBAAkB,GAAG,wBAAwB,CAAA;AAEnD,2BAA2B,SAAc,EAAE,aAA6B;IAEtE,IAAM,kBAAkB,GAAG,qDAAyB,CAAC,SAAS,EAAE,aAAa,CAAC,CAAA;IAE9E,2BAA2B,MAAc,EAAE,YAAoB;QAC7D,IAAM,OAAO,GAAG,kBAAkB,CAAiB,MAAM,YAAS,EAAE,CAAC,EAAK,YAAY,aAAU,CAAC,CAAA;QACjG,IAAM,IAAI,GAAG,kBAAkB,CAAiB,MAAM,UAAO,EAAE,CAAC,EAAK,YAAY,UAAO,CAAC,CAAA;QAEzF,OAAO,EAAE,OAAO,SAAA,EAAE,IAAI,MAAA,EAAE,CAAA;IAC1B,CAAC;IAED,yBAAyB,MAAc,EAAE,YAAoB;QAC3D,IAAM,OAAO,GAAG,kBAAkB,CAAiB,MAAM,YAAS,EAAE,CAAC,EAAK,YAAY,aAAU,CAAC,CAAA;QACjG,IAAM,IAAI,GAAG,kBAAkB,CAAiB,MAAM,UAAO,EAAE,CAAC,EAAK,YAAY,UAAO,CAAC,CAAA;QAEzF,OAAO,EAAE,OAAO,SAAA,EAAE,IAAI,MAAA,EAAE,CAAA;IAC1B,CAAC;IAED,OAAO;QACL,iBAAiB,mBAAA;QACjB,eAAe,iBAAA;KAChB,CAAA;AACH,CAAC;AAED,6BACE,GAAuB;;;;;wBAGL,qBAAM,6BAAa,CAAC,GAAG,EAAE,kBAAkB,CAAC,EAAA;;oBAAxD,SAAS,GAAG,SAA4C;oBACxD,aAAa,GAAmB,EAAE,CAAA;oBAElC,KAGF,iBAAiB,CAAC,SAAS,EAAE,aAAa,CAAC,EAF7C,iBAAiB,uBAAA,EACjB,eAAe,qBAAA,CAC8B;oBAEzC,MAAM,GAAG;wBACb,KAAK,EAAE,iBAAiB,CAAC,UAAU,EAAE,OAAO,CAAC;wBAC7C,KAAK,EAAE,iBAAiB,CAAC,UAAU,EAAE,OAAO,CAAC;wBAC7C,KAAK,EAAE,iBAAiB,CAAC,UAAU,EAAE,OAAO,CAAC;wBAC7C,KAAK,EAAE,iBAAiB,CAAC,UAAU,EAAE,OAAO,CAAC;wBAC7C,KAAK,EAAE,iBAAiB,CAAC,UAAU,EAAE,OAAO,CAAC;wBAC7C,KAAK,EAAE,iBAAiB,CAAC,UAAU,EAAE,OAAO,CAAC;wBAC7C,KAAK,EAAE,iBAAiB,CAAC,UAAU,EAAE,OAAO,CAAC;wBAC7C,KAAK,EAAE,iBAAiB,CAAC,UAAU,EAAE,OAAO,CAAC;wBAC7C,GAAG,EAAE,eAAe,CAAC,OAAO,EAAE,KAAK,CAAC;wBACpC,GAAG,EAAE,eAAe,CAAC,QAAQ,EAAE,KAAK,CAAC;qBACtC,CAAA;oBAED,uDAA0B,CAAC,SAAS,EAAE,aAAa,CAAC,CAAA;oBAEpD,sBAAO,EAAE,MAAM,QAAA,EAAE,aAAa,eAAA,EAAE,EAAA;;;;CACjC;AA5BD,kDA4BC"}
\ No newline at end of file \ No newline at end of file
build/faceLandmarkNet/types.d.ts
...@@ -5,14 +5,14 @@ export declare type FCParams = { ...@@ -5,14 +5,14 @@ export declare type FCParams = {
bias: tf.Tensor1D; bias: tf.Tensor1D;
}; };
export declare type NetParams = { export declare type NetParams = {
conv0_params: ConvParams; conv0: ConvParams;
conv1_params: ConvParams; conv1: ConvParams;
conv2_params: ConvParams; conv2: ConvParams;
conv3_params: ConvParams; conv3: ConvParams;
conv4_params: ConvParams; conv4: ConvParams;
conv5_params: ConvParams; conv5: ConvParams;
conv6_params: ConvParams; conv6: ConvParams;
conv7_params: ConvParams; conv7: ConvParams;
fc0_params: FCParams; fc0: FCParams;
fc1_params: FCParams; fc1: FCParams;
}; };
build/faceRecognitionNet/FaceRecognitionNet.d.ts
import * as tf from '@tensorflow/tfjs-core'; import * as tf from '@tensorflow/tfjs-core';
import { NeuralNetwork } from '../commons/NeuralNetwork';
import { NetInput } from '../NetInput'; import { NetInput } from '../NetInput';
import { TNetInput } from '../types'; import { TNetInput } from '../types';
export declare class FaceRecognitionNet { import { NetParams } from './types';
private _params; export declare class FaceRecognitionNet extends NeuralNetwork<NetParams> {
load(weightsOrUrl: Float32Array | string | undefined): Promise<void>; constructor();
extractWeights(weights: Float32Array): void;
forwardInput(input: NetInput): tf.Tensor2D; forwardInput(input: NetInput): tf.Tensor2D;
forward(input: TNetInput): Promise<tf.Tensor2D>; forward(input: TNetInput): Promise<tf.Tensor2D>;
computeFaceDescriptor(input: TNetInput): Promise<Float32Array | Float32Array[]>; computeFaceDescriptor(input: TNetInput): Promise<Float32Array | Float32Array[]>;
protected loadQuantizedParams(uri: string | undefined): Promise<{
params: NetParams;
paramMappings: {
originalPath?: string | undefined;
paramPath: string;
}[];
}>;
protected extractParams(weights: Float32Array): {
params: NetParams;
paramMappings: {
originalPath?: string | undefined;
paramPath: string;
}[];
};
} }
build/faceRecognitionNet/FaceRecognitionNet.js
...@@ -2,66 +2,44 @@ ...@@ -2,66 +2,44 @@
Object.defineProperty(exports, "__esModule", { value: true }); Object.defineProperty(exports, "__esModule", { value: true });
var tslib_1 = require("tslib"); var tslib_1 = require("tslib");
var tf = require("@tensorflow/tfjs-core"); var tf = require("@tensorflow/tfjs-core");
var NeuralNetwork_1 = require("../commons/NeuralNetwork");
var toNetInput_1 = require("../toNetInput"); var toNetInput_1 = require("../toNetInput");
var convLayer_1 = require("./convLayer"); var convLayer_1 = require("./convLayer");
var extractParams_1 = require("./extractParams"); var extractParams_1 = require("./extractParams");
var loadQuantizedParams_1 = require("./loadQuantizedParams"); var loadQuantizedParams_1 = require("./loadQuantizedParams");
var normalize_1 = require("./normalize"); var normalize_1 = require("./normalize");
var residualLayer_1 = require("./residualLayer"); var residualLayer_1 = require("./residualLayer");
var FaceRecognitionNet = /** @class */ (function () { var FaceRecognitionNet = /** @class */ (function (_super) {
tslib_1.__extends(FaceRecognitionNet, _super);
function FaceRecognitionNet() { function FaceRecognitionNet() {
return _super.call(this, 'FaceRecognitionNet') || this;
} }
FaceRecognitionNet.prototype.load = function (weightsOrUrl) {
return tslib_1.__awaiter(this, void 0, void 0, function () {
var _a;
return tslib_1.__generator(this, function (_b) {
switch (_b.label) {
case 0:
if (weightsOrUrl instanceof Float32Array) {
this.extractWeights(weightsOrUrl);
return [2 /*return*/];
}
if (weightsOrUrl && typeof weightsOrUrl !== 'string') {
throw new Error('FaceLandmarkNet.load - expected model uri, or weights as Float32Array');
}
_a = this;
return [4 /*yield*/, loadQuantizedParams_1.loadQuantizedParams(weightsOrUrl)];
case 1:
_a._params = _b.sent();
return [2 /*return*/];
}
});
});
};
FaceRecognitionNet.prototype.extractWeights = function (weights) {
this._params = extractParams_1.extractParams(weights);
};
FaceRecognitionNet.prototype.forwardInput = function (input) { FaceRecognitionNet.prototype.forwardInput = function (input) {
var _this = this; var params = this.params;
if (!this._params) { if (!params) {
throw new Error('FaceRecognitionNet - load model before inference'); throw new Error('FaceRecognitionNet - load model before inference');
} }
return tf.tidy(function () { return tf.tidy(function () {
var batchTensor = input.toBatchTensor(150, true); var batchTensor = input.toBatchTensor(150, true);
var normalized = normalize_1.normalize(batchTensor); var normalized = normalize_1.normalize(batchTensor);
var out = convLayer_1.convDown(normalized, _this._params.conv32_down); var out = convLayer_1.convDown(normalized, params.conv32_down);
out = tf.maxPool(out, 3, 2, 'valid'); out = tf.maxPool(out, 3, 2, 'valid');
out = residualLayer_1.residual(out, _this._params.conv32_1); out = residualLayer_1.residual(out, params.conv32_1);
out = residualLayer_1.residual(out, _this._params.conv32_2); out = residualLayer_1.residual(out, params.conv32_2);
out = residualLayer_1.residual(out, _this._params.conv32_3); out = residualLayer_1.residual(out, params.conv32_3);
out = residualLayer_1.residualDown(out, _this._params.conv64_down); out = residualLayer_1.residualDown(out, params.conv64_down);
out = residualLayer_1.residual(out, _this._params.conv64_1); out = residualLayer_1.residual(out, params.conv64_1);
out = residualLayer_1.residual(out, _this._params.conv64_2); out = residualLayer_1.residual(out, params.conv64_2);
out = residualLayer_1.residual(out, _this._params.conv64_3); out = residualLayer_1.residual(out, params.conv64_3);
out = residualLayer_1.residualDown(out, _this._params.conv128_down); out = residualLayer_1.residualDown(out, params.conv128_down);
out = residualLayer_1.residual(out, _this._params.conv128_1); out = residualLayer_1.residual(out, params.conv128_1);
out = residualLayer_1.residual(out, _this._params.conv128_2); out = residualLayer_1.residual(out, params.conv128_2);
out = residualLayer_1.residualDown(out, _this._params.conv256_down); out = residualLayer_1.residualDown(out, params.conv256_down);
out = residualLayer_1.residual(out, _this._params.conv256_1); out = residualLayer_1.residual(out, params.conv256_1);
out = residualLayer_1.residual(out, _this._params.conv256_2); out = residualLayer_1.residual(out, params.conv256_2);
out = residualLayer_1.residualDown(out, _this._params.conv256_down_out); out = residualLayer_1.residualDown(out, params.conv256_down_out);
var globalAvg = out.mean([1, 2]); var globalAvg = out.mean([1, 2]);
var fullyConnected = tf.matMul(globalAvg, _this._params.fc); var fullyConnected = tf.matMul(globalAvg, params.fc);
return fullyConnected; return fullyConnected;
}); });
}; };
...@@ -99,7 +77,13 @@ var FaceRecognitionNet = /** @class */ (function () { ...@@ -99,7 +77,13 @@ var FaceRecognitionNet = /** @class */ (function () {
}); });
}); });
}; };
FaceRecognitionNet.prototype.loadQuantizedParams = function (uri) {
return loadQuantizedParams_1.loadQuantizedParams(uri);
};
FaceRecognitionNet.prototype.extractParams = function (weights) {
return extractParams_1.extractParams(weights);
};
return FaceRecognitionNet; return FaceRecognitionNet;
}()); }(NeuralNetwork_1.NeuralNetwork));
exports.FaceRecognitionNet = FaceRecognitionNet; exports.FaceRecognitionNet = FaceRecognitionNet;
//# sourceMappingURL=FaceRecognitionNet.js.map //# sourceMappingURL=FaceRecognitionNet.js.map
\ No newline at end of file
build/faceRecognitionNet/FaceRecognitionNet.js.map
{"version":3,"file":"FaceRecognitionNet.js","sourceRoot":"","sources":["../../src/faceRecognitionNet/FaceRecognitionNet.ts"],"names":[],"mappings":";;;AAAA,0CAA4C;AAG5C,4CAA2C;AAE3C,yCAAuC;AACvC,iDAAgD;AAChD,6DAA4D;AAC5D,yCAAwC;AACxC,iDAAyD;AAGzD;IAAA;IA+EA,CAAC;IA3Ec,iCAAI,GAAjB,UAAkB,YAA+C;;;;;;wBAC/D,IAAI,YAAY,YAAY,YAAY,EAAE;4BACxC,IAAI,CAAC,cAAc,CAAC,YAAY,CAAC,CAAA;4BACjC,sBAAM;yBACP;wBAED,IAAI,YAAY,IAAI,OAAO,YAAY,KAAK,QAAQ,EAAE;4BACpD,MAAM,IAAI,KAAK,CAAC,uEAAuE,CAAC,CAAA;yBACzF;wBACD,KAAA,IAAI,CAAA;wBAAW,qBAAM,yCAAmB,CAAC,YAAY,CAAC,EAAA;;wBAAtD,GAAK,OAAO,GAAG,SAAuC,CAAA;;;;;KACvD;IAEM,2CAAc,GAArB,UAAsB,OAAqB;QACzC,IAAI,CAAC,OAAO,GAAG,6BAAa,CAAC,OAAO,CAAC,CAAA;IACvC,CAAC;IAEM,yCAAY,GAAnB,UAAoB,KAAe;QAAnC,iBAoCC;QAnCC,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE;YACjB,MAAM,IAAI,KAAK,CAAC,kDAAkD,CAAC,CAAA;SACpE;QAED,OAAO,EAAE,CAAC,IAAI,CAAC;YACb,IAAM,WAAW,GAAG,KAAK,CAAC,aAAa,CAAC,GAAG,EAAE,IAAI,CAAC,CAAA;YAElD,IAAM,UAAU,GAAG,qBAAS,CAAC,WAAW,CAAC,CAAA;YAEzC,IAAI,GAAG,GAAG,oBAAQ,CAAC,UAAU,EAAE,KAAI,CAAC,OAAO,CAAC,WAAW,CAAC,CAAA;YACxD,GAAG,GAAG,EAAE,CAAC,OAAO,CAAC,GAAG,EAAE,CAAC,EAAE,CAAC,EAAE,OAAO,CAAC,CAAA;YAEpC,GAAG,GAAG,wBAAQ,CAAC,GAAG,EAAE,KAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAA;YAC1C,GAAG,GAAG,wBAAQ,CAAC,GAAG,EAAE,KAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAA;YAC1C,GAAG,GAAG,wBAAQ,CAAC,GAAG,EAAE,KAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAA;YAE1C,GAAG,GAAG,4BAAY,CAAC,GAAG,EAAE,KAAI,CAAC,OAAO,CAAC,WAAW,CAAC,CAAA;YACjD,GAAG,GAAG,wBAAQ,CAAC,GAAG,EAAE,KAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAA;YAC1C,GAAG,GAAG,wBAAQ,CAAC,GAAG,EAAE,KAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAA;YAC1C,GAAG,GAAG,wBAAQ,CAAC,GAAG,EAAE,KAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAA;YAE1C,GAAG,GAAG,4BAAY,CAAC,GAAG,EAAE,KAAI,CAAC,OAAO,CAAC,YAAY,CAAC,CAAA;YAClD,GAAG,GAAG,wBAAQ,CAAC,GAAG,EAAE,KAAI,CAAC,OAAO,CAAC,SAAS,CAAC,CAAA;YAC3C,GAAG,GAAG,wBAAQ,CAAC,GAAG,EAAE,KAAI,CAAC,OAAO,CAAC,SAAS,CAAC,CAAA;YAE3C,GAAG,GAAG,4BAAY,CAAC,GAAG,EAAE,KAAI,CAAC,OAAO,CAAC,YAAY,CAAC,CAAA;YAClD,GAAG,GAAG,wBAAQ,CAAC,GAAG,EAAE,KAAI,CAAC,OAAO,CAAC,SAAS,CAAC,CAAA;YAC3C,GAAG,GAAG,wBAAQ,CAAC,GAAG,EAAE,KAAI,CAAC,OAAO,CAAC,SAAS,CAAC,CAAA;YAC3C,GAAG,GAAG,4BAAY,CAAC,GAAG,EAAE,KAAI,CAAC,OAAO,CAAC,gBAAgB,CAAC,CAAA;YAEtD,IAAM,SAAS,GAAG,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAgB,CAAA;YACjD,IAAM,cAAc,GAAG,EAAE,CAAC,MAAM,CAAC,SAAS,EAAE,KAAI,CAAC,OAAO,CAAC,EAAE,CAAC,CAAA;YAE5D,OAAO,cAAc,CAAA;QACvB,CAAC,CAAC,CAAA;IACJ,CAAC;IAEY,oCAAO,GAApB,UAAqB,KAAgB;;;;;;wBAC5B,KAAA,IAAI,CAAC,YAAY,CAAA;wBAAC,qBAAM,uBAAU,CAAC,KAAK,EAAE,IAAI,CAAC,EAAA;4BAAtD,sBAAO,SAAA,IAAI,GAAc,SAA6B,EAAC,EAAA;;;;KACxD;IAEY,kDAAqB,GAAlC,UAAmC,KAAgB;;;;;;4BAChC,qBAAM,uBAAU,CAAC,KAAK,EAAE,IAAI,CAAC,EAAA;;wBAAxC,QAAQ,GAAG,SAA6B;wBAExC,qBAAqB,GAAG,EAAE,CAAC,IAAI,CACnC,cAAM,OAAA,EAAE,CAAC,OAAO,CAAC,KAAI,CAAC,YAAY,CAAC,QAAQ,CAAC,CAAC,EAAvC,CAAuC,CAC9C,CAAA;wBAE+B,qBAAM,OAAO,CAAC,GAAG,CAAC,qBAAqB,CAAC,GAAG,CACzE,UAAA,CAAC,IAAI,OAAA,CAAC,CAAC,IAAI,EAAE,EAAR,CAAQ,CACd,CAAC,EAAA;;wBAFI,uBAAuB,GAAG,SAEZ;wBAEpB,qBAAqB,CAAC,OAAO,CAAC,UAAA,CAAC,IAAI,OAAA,CAAC,CAAC,OAAO,EAAE,EAAX,CAAW,CAAC,CAAA;wBAE/C,sBAAO,QAAQ,CAAC,YAAY;gCAC1B,CAAC,CAAC,uBAAuB;gCACzB,CAAC,CAAC,uBAAuB,CAAC,CAAC,CAAC,EAAA;;;;KAC/B;IACH,yBAAC;AAAD,CAAC,AA/ED,IA+EC;AA/EY,gDAAkB"} {"version":3,"file":"FaceRecognitionNet.js","sourceRoot":"","sources":["../../src/faceRecognitionNet/FaceRecognitionNet.ts"],"names":[],"mappings":";;;AAAA,0CAA4C;AAE5C,0DAAyD;AAEzD,4CAA2C;AAE3C,yCAAuC;AACvC,iDAAgD;AAChD,6DAA4D;AAC5D,yCAAwC;AACxC,iDAAyD;AAGzD;IAAwC,8CAAwB;IAE9D;eACE,kBAAM,oBAAoB,CAAC;IAC7B,CAAC;IAEM,yCAAY,GAAnB,UAAoB,KAAe;QAEzB,IAAA,oBAAM,CAAS;QAEvB,IAAI,CAAC,MAAM,EAAE;YACX,MAAM,IAAI,KAAK,CAAC,kDAAkD,CAAC,CAAA;SACpE;QAED,OAAO,EAAE,CAAC,IAAI,CAAC;YACb,IAAM,WAAW,GAAG,KAAK,CAAC,aAAa,CAAC,GAAG,EAAE,IAAI,CAAC,CAAA;YAElD,IAAM,UAAU,GAAG,qBAAS,CAAC,WAAW,CAAC,CAAA;YAEzC,IAAI,GAAG,GAAG,oBAAQ,CAAC,UAAU,EAAE,MAAM,CAAC,WAAW,CAAC,CAAA;YAClD,GAAG,GAAG,EAAE,CAAC,OAAO,CAAC,GAAG,EAAE,CAAC,EAAE,CAAC,EAAE,OAAO,CAAC,CAAA;YAEpC,GAAG,GAAG,wBAAQ,CAAC,GAAG,EAAE,MAAM,CAAC,QAAQ,CAAC,CAAA;YACpC,GAAG,GAAG,wBAAQ,CAAC,GAAG,EAAE,MAAM,CAAC,QAAQ,CAAC,CAAA;YACpC,GAAG,GAAG,wBAAQ,CAAC,GAAG,EAAE,MAAM,CAAC,QAAQ,CAAC,CAAA;YAEpC,GAAG,GAAG,4BAAY,CAAC,GAAG,EAAE,MAAM,CAAC,WAAW,CAAC,CAAA;YAC3C,GAAG,GAAG,wBAAQ,CAAC,GAAG,EAAE,MAAM,CAAC,QAAQ,CAAC,CAAA;YACpC,GAAG,GAAG,wBAAQ,CAAC,GAAG,EAAE,MAAM,CAAC,QAAQ,CAAC,CAAA;YACpC,GAAG,GAAG,wBAAQ,CAAC,GAAG,EAAE,MAAM,CAAC,QAAQ,CAAC,CAAA;YAEpC,GAAG,GAAG,4BAAY,CAAC,GAAG,EAAE,MAAM,CAAC,YAAY,CAAC,CAAA;YAC5C,GAAG,GAAG,wBAAQ,CAAC,GAAG,EAAE,MAAM,CAAC,SAAS,CAAC,CAAA;YACrC,GAAG,GAAG,wBAAQ,CAAC,GAAG,EAAE,MAAM,CAAC,SAAS,CAAC,CAAA;YAErC,GAAG,GAAG,4BAAY,CAAC,GAAG,EAAE,MAAM,CAAC,YAAY,CAAC,CAAA;YAC5C,GAAG,GAAG,wBAAQ,CAAC,GAAG,EAAE,MAAM,CAAC,SAAS,CAAC,CAAA;YACrC,GAAG,GAAG,wBAAQ,CAAC,GAAG,EAAE,MAAM,CAAC,SAAS,CAAC,CAAA;YACrC,GAAG,GAAG,4BAAY,CAAC,GAAG,EAAE,MAAM,CAAC,gBAAgB,CAAC,CAAA;YAEhD,IAAM,SAAS,GAAG,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAgB,CAAA;YACjD,IAAM,cAAc,GAAG,EAAE,CAAC,MAAM,CAAC,SAAS,EAAE,MAAM,CAAC,EAAE,CAAC,CAAA;YAEtD,OAAO,cAAc,CAAA;QACvB,CAAC,CAAC,CAAA;IACJ,CAAC;IAEY,oCAAO,GAApB,UAAqB,KAAgB;;;;;;wBAC5B,KAAA,IAAI,CAAC,YAAY,CAAA;wBAAC,qBAAM,uBAAU,CAAC,KAAK,EAAE,IAAI,CAAC,EAAA;4BAAtD,sBAAO,SAAA,IAAI,GAAc,SAA6B,EAAC,EAAA;;;;KACxD;IAEY,kDAAqB,GAAlC,UAAmC,KAAgB;;;;;;4BAChC,qBAAM,uBAAU,CAAC,KAAK,EAAE,IAAI,CAAC,EAAA;;wBAAxC,QAAQ,GAAG,SAA6B;wBAExC,qBAAqB,GAAG,EAAE,CAAC,IAAI,CACnC,cAAM,OAAA,EAAE,CAAC,OAAO,CAAC,KAAI,CAAC,YAAY,CAAC,QAAQ,CAAC,CAAC,EAAvC,CAAuC,CAC9C,CAAA;wBAE+B,qBAAM,OAAO,CAAC,GAAG,CAAC,qBAAqB,CAAC,GAAG,CACzE,UAAA,CAAC,IAAI,OAAA,CAAC,CAAC,IAAI,EAAE,EAAR,CAAQ,CACd,CAAC,EAAA;;wBAFI,uBAAuB,GAAG,SAEZ;wBAEpB,qBAAqB,CAAC,OAAO,CAAC,UAAA,CAAC,IAAI,OAAA,CAAC,CAAC,OAAO,EAAE,EAAX,CAAW,CAAC,CAAA;wBAE/C,sBAAO,QAAQ,CAAC,YAAY;gCAC1B,CAAC,CAAC,uBAAuB;gCACzB,CAAC,CAAC,uBAAuB,CAAC,CAAC,CAAC,EAAA;;;;KAC/B;IAES,gDAAmB,GAA7B,UAA8B,GAAuB;QACnD,OAAO,yCAAmB,CAAC,GAAG,CAAC,CAAA;IACjC,CAAC;IAES,0CAAa,GAAvB,UAAwB,OAAqB;QAC3C,OAAO,6BAAa,CAAC,OAAO,CAAC,CAAA;IAC/B,CAAC;IACH,yBAAC;AAAD,CAAC,AA5ED,CAAwC,6BAAa,GA4EpD;AA5EY,gDAAkB"}
\ No newline at end of file \ No newline at end of file
build/faceRecognitionNet/extractParams.d.ts
import { ParamMapping } from '../commons/types';
import { NetParams } from './types'; import { NetParams } from './types';
export declare function extractParams(weights: Float32Array): NetParams; export declare function extractParams(weights: Float32Array): {
params: NetParams;
paramMappings: ParamMapping[];
};
build/faceRecognitionNet/extractParams.js
...@@ -3,43 +3,40 @@ Object.defineProperty(exports, "__esModule", { value: true }); ...@@ -3,43 +3,40 @@ Object.defineProperty(exports, "__esModule", { value: true });
var tf = require("@tensorflow/tfjs-core"); var tf = require("@tensorflow/tfjs-core");
var extractWeightsFactory_1 = require("../commons/extractWeightsFactory"); var extractWeightsFactory_1 = require("../commons/extractWeightsFactory");
var utils_1 = require("../utils"); var utils_1 = require("../utils");
function extractorsFactory(extractWeights) { function extractorsFactory(extractWeights, paramMappings) {
function extractFilterValues(numFilterValues, numFilters, filterSize) { function extractFilterValues(numFilterValues, numFilters, filterSize) {
var weights = extractWeights(numFilterValues); var weights = extractWeights(numFilterValues);
var depth = weights.length / (numFilters * filterSize * filterSize); var depth = weights.length / (numFilters * filterSize * filterSize);
if (utils_1.isFloat(depth)) { if (utils_1.isFloat(depth)) {
throw new Error("depth has to be an integer: " + depth + ", weights.length: " + weights.length + ", numFilters: " + numFilters + ", filterSize: " + filterSize); throw new Error("depth has to be an integer: " + depth + ", weights.length: " + weights.length + ", numFilters: " + numFilters + ", filterSize: " + filterSize);
} }
return tf.transpose(tf.tensor4d(weights, [numFilters, depth, filterSize, filterSize]), [2, 3, 1, 0]); return tf.tidy(function () { return tf.transpose(tf.tensor4d(weights, [numFilters, depth, filterSize, filterSize]), [2, 3, 1, 0]); });
} }
function extractScaleLayerParams(numWeights) { function extractConvParams(numFilterValues, numFilters, filterSize, mappedPrefix) {
var filters = extractFilterValues(numFilterValues, numFilters, filterSize);
var bias = tf.tensor1d(extractWeights(numFilters));
paramMappings.push({ paramPath: mappedPrefix + "/filters" }, { paramPath: mappedPrefix + "/bias" });
return { filters: filters, bias: bias };
}
function extractScaleLayerParams(numWeights, mappedPrefix) {
var weights = tf.tensor1d(extractWeights(numWeights)); var weights = tf.tensor1d(extractWeights(numWeights));
var biases = tf.tensor1d(extractWeights(numWeights)); var biases = tf.tensor1d(extractWeights(numWeights));
paramMappings.push({ paramPath: mappedPrefix + "/weights" }, { paramPath: mappedPrefix + "/biases" });
return { return {
weights: weights, weights: weights,
biases: biases biases: biases
}; };
} }
function extractConvLayerParams(numFilterValues, numFilters, filterSize) { function extractConvLayerParams(numFilterValues, numFilters, filterSize, mappedPrefix) {
var conv_filters = extractFilterValues(numFilterValues, numFilters, filterSize); var conv = extractConvParams(numFilterValues, numFilters, filterSize, mappedPrefix + "/conv");
var conv_bias = tf.tensor1d(extractWeights(numFilters)); var scale = extractScaleLayerParams(numFilters, mappedPrefix + "/scale");
var scale = extractScaleLayerParams(numFilters); return { conv: conv, scale: scale };
return {
conv: {
filters: conv_filters,
bias: conv_bias
},
scale: scale
};
} }
function extractResidualLayerParams(numFilterValues, numFilters, filterSize, isDown) { function extractResidualLayerParams(numFilterValues, numFilters, filterSize, mappedPrefix, isDown) {
if (isDown === void 0) { isDown = false; } if (isDown === void 0) { isDown = false; }
var conv1 = extractConvLayerParams((isDown ? 0.5 : 1) * numFilterValues, numFilters, filterSize); var conv1 = extractConvLayerParams((isDown ? 0.5 : 1) * numFilterValues, numFilters, filterSize, mappedPrefix + "/conv1");
var conv2 = extractConvLayerParams(numFilterValues, numFilters, filterSize); var conv2 = extractConvLayerParams(numFilterValues, numFilters, filterSize, mappedPrefix + "/conv2");
return { return { conv1: conv1, conv2: conv2 };
conv1: conv1,
conv2: conv2
};
} }
return { return {
extractConvLayerParams: extractConvLayerParams, extractConvLayerParams: extractConvLayerParams,
...@@ -48,27 +45,29 @@ function extractorsFactory(extractWeights) { ...@@ -48,27 +45,29 @@ function extractorsFactory(extractWeights) {
} }
function extractParams(weights) { function extractParams(weights) {
var _a = extractWeightsFactory_1.extractWeightsFactory(weights), extractWeights = _a.extractWeights, getRemainingWeights = _a.getRemainingWeights; var _a = extractWeightsFactory_1.extractWeightsFactory(weights), extractWeights = _a.extractWeights, getRemainingWeights = _a.getRemainingWeights;
var _b = extractorsFactory(extractWeights), extractConvLayerParams = _b.extractConvLayerParams, extractResidualLayerParams = _b.extractResidualLayerParams; var paramMappings = [];
var conv32_down = extractConvLayerParams(4704, 32, 7); var _b = extractorsFactory(extractWeights, paramMappings), extractConvLayerParams = _b.extractConvLayerParams, extractResidualLayerParams = _b.extractResidualLayerParams;
var conv32_1 = extractResidualLayerParams(9216, 32, 3); var conv32_down = extractConvLayerParams(4704, 32, 7, 'conv32_down');
var conv32_2 = extractResidualLayerParams(9216, 32, 3); var conv32_1 = extractResidualLayerParams(9216, 32, 3, 'conv32_1');
var conv32_3 = extractResidualLayerParams(9216, 32, 3); var conv32_2 = extractResidualLayerParams(9216, 32, 3, 'conv32_2');
var conv64_down = extractResidualLayerParams(36864, 64, 3, true); var conv32_3 = extractResidualLayerParams(9216, 32, 3, 'conv32_3');
var conv64_1 = extractResidualLayerParams(36864, 64, 3); var conv64_down = extractResidualLayerParams(36864, 64, 3, 'conv64_down', true);
var conv64_2 = extractResidualLayerParams(36864, 64, 3); var conv64_1 = extractResidualLayerParams(36864, 64, 3, 'conv64_1');
var conv64_3 = extractResidualLayerParams(36864, 64, 3); var conv64_2 = extractResidualLayerParams(36864, 64, 3, 'conv64_2');
var conv128_down = extractResidualLayerParams(147456, 128, 3, true); var conv64_3 = extractResidualLayerParams(36864, 64, 3, 'conv64_3');
var conv128_1 = extractResidualLayerParams(147456, 128, 3); var conv128_down = extractResidualLayerParams(147456, 128, 3, 'conv128_down', true);
var conv128_2 = extractResidualLayerParams(147456, 128, 3); var conv128_1 = extractResidualLayerParams(147456, 128, 3, 'conv128_1');
var conv256_down = extractResidualLayerParams(589824, 256, 3, true); var conv128_2 = extractResidualLayerParams(147456, 128, 3, 'conv128_2');
var conv256_1 = extractResidualLayerParams(589824, 256, 3); var conv256_down = extractResidualLayerParams(589824, 256, 3, 'conv256_down', true);
var conv256_2 = extractResidualLayerParams(589824, 256, 3); var conv256_1 = extractResidualLayerParams(589824, 256, 3, 'conv256_1');
var conv256_down_out = extractResidualLayerParams(589824, 256, 3); var conv256_2 = extractResidualLayerParams(589824, 256, 3, 'conv256_2');
var fc = tf.transpose(tf.tensor2d(extractWeights(256 * 128), [128, 256]), [1, 0]); var conv256_down_out = extractResidualLayerParams(589824, 256, 3, 'conv256_down_out');
var fc = tf.tidy(function () { return tf.transpose(tf.tensor2d(extractWeights(256 * 128), [128, 256]), [1, 0]); });
paramMappings.push({ paramPath: "fc" });
if (getRemainingWeights().length !== 0) { if (getRemainingWeights().length !== 0) {
throw new Error("weights remaing after extract: " + getRemainingWeights().length); throw new Error("weights remaing after extract: " + getRemainingWeights().length);
} }
return { var params = {
conv32_down: conv32_down, conv32_down: conv32_down,
conv32_1: conv32_1, conv32_1: conv32_1,
conv32_2: conv32_2, conv32_2: conv32_2,
...@@ -86,6 +85,7 @@ function extractParams(weights) { ...@@ -86,6 +85,7 @@ function extractParams(weights) {
conv256_down_out: conv256_down_out, conv256_down_out: conv256_down_out,
fc: fc fc: fc
}; };
return { params: params, paramMappings: paramMappings };
} }
exports.extractParams = extractParams; exports.extractParams = extractParams;
//# sourceMappingURL=extractParams.js.map //# sourceMappingURL=extractParams.js.map
\ No newline at end of file
build/faceRecognitionNet/extractParams.js.map
{"version":3,"file":"extractParams.js","sourceRoot":"","sources":["../../src/faceRecognitionNet/extractParams.ts"],"names":[],"mappings":";;AAAA,0CAA4C;AAE5C,0EAAyE;AAEzE,kCAAmC;AAGnC,2BAA2B,cAAsC;IAE/D,6BAA6B,eAAuB,EAAE,UAAkB,EAAE,UAAkB;QAC1F,IAAM,OAAO,GAAG,cAAc,CAAC,eAAe,CAAC,CAAA;QAC/C,IAAM,KAAK,GAAG,OAAO,CAAC,MAAM,GAAG,CAAC,UAAU,GAAG,UAAU,GAAG,UAAU,CAAC,CAAA;QAErE,IAAI,eAAO,CAAC,KAAK,CAAC,EAAE;YAClB,MAAM,IAAI,KAAK,CAAC,iCAA+B,KAAK,0BAAqB,OAAO,CAAC,MAAM,sBAAiB,UAAU,sBAAiB,UAAY,CAAC,CAAA;SACjJ;QAED,OAAO,EAAE,CAAC,SAAS,CACjB,EAAE,CAAC,QAAQ,CAAC,OAAO,EAAE,CAAC,UAAU,EAAE,KAAK,EAAE,UAAU,EAAE,UAAU,CAAC,CAAC,EACjE,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,CACb,CAAA;IACH,CAAC;IAED,iCAAiC,UAAkB;QACjD,IAAM,OAAO,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,UAAU,CAAC,CAAC,CAAA;QACvD,IAAM,MAAM,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,UAAU,CAAC,CAAC,CAAA;QACtD,OAAO;YACL,OAAO,SAAA;YACP,MAAM,QAAA;SACP,CAAA;IACH,CAAC;IAED,gCACE,eAAuB,EACvB,UAAkB,EAClB,UAAkB;QAElB,IAAM,YAAY,GAAG,mBAAmB,CAAC,eAAe,EAAE,UAAU,EAAE,UAAU,CAAC,CAAA;QACjF,IAAM,SAAS,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,UAAU,CAAC,CAAC,CAAA;QACzD,IAAM,KAAK,GAAG,uBAAuB,CAAC,UAAU,CAAC,CAAA;QAEjD,OAAO;YACL,IAAI,EAAE;gBACJ,OAAO,EAAE,YAAY;gBACrB,IAAI,EAAE,SAAS;aAChB;YACD,KAAK,OAAA;SACN,CAAA;IACH,CAAC;IAED,oCACE,eAAuB,EACvB,UAAkB,EAClB,UAAkB,EAClB,MAAuB;QAAvB,uBAAA,EAAA,cAAuB;QAEvB,IAAM,KAAK,GAAoB,sBAAsB,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,eAAe,EAAE,UAAU,EAAE,UAAU,CAAC,CAAA;QACnH,IAAM,KAAK,GAAoB,sBAAsB,CAAC,eAAe,EAAE,UAAU,EAAE,UAAU,CAAC,CAAA;QAE9F,OAAO;YACL,KAAK,OAAA;YACL,KAAK,OAAA;SACN,CAAA;IACH,CAAC;IAED,OAAO;QACL,sBAAsB,wBAAA;QACtB,0BAA0B,4BAAA;KAC3B,CAAA;AAEH,CAAC;AAED,uBAA8B,OAAqB;IAC3C,IAAA,2DAG4B,EAFhC,kCAAc,EACd,4CAAmB,CACa;IAE5B,IAAA,sCAG+B,EAFnC,kDAAsB,EACtB,0DAA0B,CACS;IAErC,IAAM,WAAW,GAAG,sBAAsB,CAAC,IAAI,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;IACvD,IAAM,QAAQ,GAAG,0BAA0B,CAAC,IAAI,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;IACxD,IAAM,QAAQ,GAAG,0BAA0B,CAAC,IAAI,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;IACxD,IAAM,QAAQ,GAAG,0BAA0B,CAAC,IAAI,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;IAExD,IAAM,WAAW,GAAG,0BAA0B,CAAC,KAAK,EAAE,EAAE,EAAE,CAAC,EAAE,IAAI,CAAC,CAAA;IAClE,IAAM,QAAQ,GAAG,0BAA0B,CAAC,KAAK,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;IACzD,IAAM,QAAQ,GAAG,0BAA0B,CAAC,KAAK,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;IACzD,IAAM,QAAQ,GAAG,0BAA0B,CAAC,KAAK,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;IAEzD,IAAM,YAAY,GAAG,0BAA0B,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,EAAE,IAAI,CAAC,CAAA;IACrE,IAAM,SAAS,GAAG,0BAA0B,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,CAAC,CAAA;IAC5D,IAAM,SAAS,GAAG,0BAA0B,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,CAAC,CAAA;IAE5D,IAAM,YAAY,GAAG,0BAA0B,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,EAAE,IAAI,CAAC,CAAA;IACrE,IAAM,SAAS,GAAG,0BAA0B,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,CAAC,CAAA;IAC5D,IAAM,SAAS,GAAG,0BAA0B,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,CAAC,CAAA;IAC5D,IAAM,gBAAgB,GAAG,0BAA0B,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,CAAC,CAAA;IAEnE,IAAM,EAAE,GAAG,EAAE,CAAC,SAAS,CAAC,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,GAAG,GAAG,GAAG,CAAC,EAAE,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAEnF,IAAI,mBAAmB,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE;QACtC,MAAM,IAAI,KAAK,CAAC,oCAAkC,mBAAmB,EAAE,CAAC,MAAQ,CAAC,CAAA;KAClF;IAED,OAAO;QACL,WAAW,aAAA;QACX,QAAQ,UAAA;QACR,QAAQ,UAAA;QACR,QAAQ,UAAA;QACR,WAAW,aAAA;QACX,QAAQ,UAAA;QACR,QAAQ,UAAA;QACR,QAAQ,UAAA;QACR,YAAY,cAAA;QACZ,SAAS,WAAA;QACT,SAAS,WAAA;QACT,YAAY,cAAA;QACZ,SAAS,WAAA;QACT,SAAS,WAAA;QACT,gBAAgB,kBAAA;QAChB,EAAE,IAAA;KACH,CAAA;AACH,CAAC;AAtDD,sCAsDC"} {"version":3,"file":"extractParams.js","sourceRoot":"","sources":["../../src/faceRecognitionNet/extractParams.ts"],"names":[],"mappings":";;AAAA,0CAA4C;AAE5C,0EAAyE;AAEzE,kCAAmC;AAGnC,2BAA2B,cAAsC,EAAE,aAA6B;IAE9F,6BAA6B,eAAuB,EAAE,UAAkB,EAAE,UAAkB;QAC1F,IAAM,OAAO,GAAG,cAAc,CAAC,eAAe,CAAC,CAAA;QAC/C,IAAM,KAAK,GAAG,OAAO,CAAC,MAAM,GAAG,CAAC,UAAU,GAAG,UAAU,GAAG,UAAU,CAAC,CAAA;QAErE,IAAI,eAAO,CAAC,KAAK,CAAC,EAAE;YAClB,MAAM,IAAI,KAAK,CAAC,iCAA+B,KAAK,0BAAqB,OAAO,CAAC,MAAM,sBAAiB,UAAU,sBAAiB,UAAY,CAAC,CAAA;SACjJ;QAED,OAAO,EAAE,CAAC,IAAI,CACZ,cAAM,OAAA,EAAE,CAAC,SAAS,CAChB,EAAE,CAAC,QAAQ,CAAC,OAAO,EAAE,CAAC,UAAU,EAAE,KAAK,EAAE,UAAU,EAAE,UAAU,CAAC,CAAC,EACjE,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,CACb,EAHK,CAGL,CACF,CAAA;IACH,CAAC;IAED,2BACE,eAAuB,EACvB,UAAkB,EAClB,UAAkB,EAClB,YAAoB;QAGpB,IAAM,OAAO,GAAG,mBAAmB,CAAC,eAAe,EAAE,UAAU,EAAE,UAAU,CAAC,CAAA;QAC5E,IAAM,IAAI,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,UAAU,CAAC,CAAC,CAAA;QAEpD,aAAa,CAAC,IAAI,CAChB,EAAE,SAAS,EAAK,YAAY,aAAU,EAAE,EACxC,EAAE,SAAS,EAAK,YAAY,UAAO,EAAE,CACtC,CAAA;QAED,OAAO,EAAE,OAAO,SAAA,EAAE,IAAI,MAAA,EAAE,CAAA;IAC1B,CAAC;IAED,iCAAiC,UAAkB,EAAE,YAAoB;QAEvE,IAAM,OAAO,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,UAAU,CAAC,CAAC,CAAA;QACvD,IAAM,MAAM,GAAG,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,UAAU,CAAC,CAAC,CAAA;QAEtD,aAAa,CAAC,IAAI,CAChB,EAAE,SAAS,EAAK,YAAY,aAAU,EAAE,EACxC,EAAE,SAAS,EAAK,YAAY,YAAS,EAAE,CACxC,CAAA;QAED,OAAO;YACL,OAAO,SAAA;YACP,MAAM,QAAA;SACP,CAAA;IACH,CAAC;IAED,gCACE,eAAuB,EACvB,UAAkB,EAClB,UAAkB,EAClB,YAAoB;QAGpB,IAAM,IAAI,GAAG,iBAAiB,CAAC,eAAe,EAAE,UAAU,EAAE,UAAU,EAAK,YAAY,UAAO,CAAC,CAAA;QAC/F,IAAM,KAAK,GAAG,uBAAuB,CAAC,UAAU,EAAK,YAAY,WAAQ,CAAC,CAAA;QAE1E,OAAO,EAAE,IAAI,MAAA,EAAE,KAAK,OAAA,EAAE,CAAA;IACxB,CAAC;IAED,oCACE,eAAuB,EACvB,UAAkB,EAClB,UAAkB,EAClB,YAAoB,EACpB,MAAuB;QAAvB,uBAAA,EAAA,cAAuB;QAGvB,IAAM,KAAK,GAAG,sBAAsB,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,eAAe,EAAE,UAAU,EAAE,UAAU,EAAK,YAAY,WAAQ,CAAC,CAAA;QAC3H,IAAM,KAAK,GAAG,sBAAsB,CAAC,eAAe,EAAE,UAAU,EAAE,UAAU,EAAK,YAAY,WAAQ,CAAC,CAAA;QAEtG,OAAO,EAAE,KAAK,OAAA,EAAE,KAAK,OAAA,EAAE,CAAA;IACzB,CAAC;IAED,OAAO;QACL,sBAAsB,wBAAA;QACtB,0BAA0B,4BAAA;KAC3B,CAAA;AAEH,CAAC;AAED,uBAA8B,OAAqB;IAE3C,IAAA,2DAG4B,EAFhC,kCAAc,EACd,4CAAmB,CACa;IAElC,IAAM,aAAa,GAAmB,EAAE,CAAA;IAElC,IAAA,qDAG8C,EAFlD,kDAAsB,EACtB,0DAA0B,CACwB;IAEpD,IAAM,WAAW,GAAG,sBAAsB,CAAC,IAAI,EAAE,EAAE,EAAE,CAAC,EAAE,aAAa,CAAC,CAAA;IACtE,IAAM,QAAQ,GAAG,0BAA0B,CAAC,IAAI,EAAE,EAAE,EAAE,CAAC,EAAE,UAAU,CAAC,CAAA;IACpE,IAAM,QAAQ,GAAG,0BAA0B,CAAC,IAAI,EAAE,EAAE,EAAE,CAAC,EAAE,UAAU,CAAC,CAAA;IACpE,IAAM,QAAQ,GAAG,0BAA0B,CAAC,IAAI,EAAE,EAAE,EAAE,CAAC,EAAE,UAAU,CAAC,CAAA;IAEpE,IAAM,WAAW,GAAG,0BAA0B,CAAC,KAAK,EAAE,EAAE,EAAE,CAAC,EAAE,aAAa,EAAE,IAAI,CAAC,CAAA;IACjF,IAAM,QAAQ,GAAG,0BAA0B,CAAC,KAAK,EAAE,EAAE,EAAE,CAAC,EAAE,UAAU,CAAC,CAAA;IACrE,IAAM,QAAQ,GAAG,0BAA0B,CAAC,KAAK,EAAE,EAAE,EAAE,CAAC,EAAE,UAAU,CAAC,CAAA;IACrE,IAAM,QAAQ,GAAG,0BAA0B,CAAC,KAAK,EAAE,EAAE,EAAE,CAAC,EAAE,UAAU,CAAC,CAAA;IAErE,IAAM,YAAY,GAAG,0BAA0B,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,EAAE,cAAc,EAAE,IAAI,CAAC,CAAA;IACrF,IAAM,SAAS,GAAG,0BAA0B,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,EAAE,WAAW,CAAC,CAAA;IACzE,IAAM,SAAS,GAAG,0BAA0B,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,EAAE,WAAW,CAAC,CAAA;IAEzE,IAAM,YAAY,GAAG,0BAA0B,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,EAAE,cAAc,EAAE,IAAI,CAAC,CAAA;IACrF,IAAM,SAAS,GAAG,0BAA0B,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,EAAE,WAAW,CAAC,CAAA;IACzE,IAAM,SAAS,GAAG,0BAA0B,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,EAAE,WAAW,CAAC,CAAA;IACzE,IAAM,gBAAgB,GAAG,0BAA0B,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,EAAE,kBAAkB,CAAC,CAAA;IAEvF,IAAM,EAAE,GAAG,EAAE,CAAC,IAAI,CAChB,cAAM,OAAA,EAAE,CAAC,SAAS,CAAC,EAAE,CAAC,QAAQ,CAAC,cAAc,CAAC,GAAG,GAAG,GAAG,CAAC,EAAE,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAxE,CAAwE,CAC/E,CAAA;IACD,aAAa,CAAC,IAAI,CAAC,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;IAEvC,IAAI,mBAAmB,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE;QACtC,MAAM,IAAI,KAAK,CAAC,oCAAkC,mBAAmB,EAAE,CAAC,MAAQ,CAAC,CAAA;KAClF;IAED,IAAM,MAAM,GAAG;QACb,WAAW,aAAA;QACX,QAAQ,UAAA;QACR,QAAQ,UAAA;QACR,QAAQ,UAAA;QACR,WAAW,aAAA;QACX,QAAQ,UAAA;QACR,QAAQ,UAAA;QACR,QAAQ,UAAA;QACR,YAAY,cAAA;QACZ,SAAS,WAAA;QACT,SAAS,WAAA;QACT,YAAY,cAAA;QACZ,SAAS,WAAA;QACT,SAAS,WAAA;QACT,gBAAgB,kBAAA;QAChB,EAAE,IAAA;KACH,CAAA;IAED,OAAO,EAAE,MAAM,QAAA,EAAE,aAAa,eAAA,EAAE,CAAA;AAClC,CAAC;AA9DD,sCA8DC"}
\ No newline at end of file \ No newline at end of file
build/faceRecognitionNet/loadQuantizedParams.d.ts
export declare function loadQuantizedParams(uri: string | undefined): Promise<any>; import { ParamMapping } from '../commons/types';
import { NetParams } from './types';
export declare function loadQuantizedParams(uri: string | undefined): Promise<{
params: NetParams;
paramMappings: ParamMapping[];
}>;
build/faceRecognitionNet/loadQuantizedParams.js
"use strict"; "use strict";
Object.defineProperty(exports, "__esModule", { value: true }); Object.defineProperty(exports, "__esModule", { value: true });
var tslib_1 = require("tslib"); var tslib_1 = require("tslib");
var disposeUnusedWeightTensors_1 = require("../commons/disposeUnusedWeightTensors");
var extractWeightEntryFactory_1 = require("../commons/extractWeightEntryFactory");
var isTensor_1 = require("../commons/isTensor"); var isTensor_1 = require("../commons/isTensor");
var loadWeightMap_1 = require("../commons/loadWeightMap"); var loadWeightMap_1 = require("../commons/loadWeightMap");
var DEFAULT_MODEL_NAME = 'face_recognition_model'; var DEFAULT_MODEL_NAME = 'face_recognition_model';
function extractorsFactory(weightMap) { function extractorsFactory(weightMap, paramMappings) {
var extractWeightEntry = extractWeightEntryFactory_1.extractWeightEntryFactory(weightMap, paramMappings);
function extractScaleLayerParams(prefix) { function extractScaleLayerParams(prefix) {
var params = { var weights = extractWeightEntry(prefix + "/scale/weights", 1);
weights: weightMap[prefix + "/scale/weights"], var biases = extractWeightEntry(prefix + "/scale/biases", 1);
biases: weightMap[prefix + "/scale/biases"] return { weights: weights, biases: biases };
};
if (!isTensor_1.isTensor1D(params.weights)) {
throw new Error("expected weightMap[" + prefix + "/scale/weights] to be a Tensor1D, instead have " + params.weights);
}
if (!isTensor_1.isTensor1D(params.biases)) {
throw new Error("expected weightMap[" + prefix + "/scale/biases] to be a Tensor1D, instead have " + params.biases);
}
return params;
} }
function extractConvLayerParams(prefix) { function extractConvLayerParams(prefix) {
var params = { var filters = extractWeightEntry(prefix + "/conv/filters", 4);
filters: weightMap[prefix + "/conv/filters"], var bias = extractWeightEntry(prefix + "/conv/bias", 1);
bias: weightMap[prefix + "/conv/bias"] var scale = extractScaleLayerParams(prefix);
}; return { conv: { filters: filters, bias: bias }, scale: scale };
if (!isTensor_1.isTensor4D(params.filters)) {
throw new Error("expected weightMap[" + prefix + "/conv/filters] to be a Tensor1D, instead have " + params.filters);
}
if (!isTensor_1.isTensor1D(params.bias)) {
throw new Error("expected weightMap[" + prefix + "/conv/bias] to be a Tensor1D, instead have " + params.bias);
}
return {
conv: params,
scale: extractScaleLayerParams(prefix)
};
} }
function extractResidualLayerParams(prefix) { function extractResidualLayerParams(prefix) {
return { return {
...@@ -47,13 +32,14 @@ function extractorsFactory(weightMap) { ...@@ -47,13 +32,14 @@ function extractorsFactory(weightMap) {
} }
function loadQuantizedParams(uri) { function loadQuantizedParams(uri) {
return tslib_1.__awaiter(this, void 0, void 0, function () { return tslib_1.__awaiter(this, void 0, void 0, function () {
var weightMap, _a, extractConvLayerParams, extractResidualLayerParams, conv32_down, conv32_1, conv32_2, conv32_3, conv64_down, conv64_1, conv64_2, conv64_3, conv128_down, conv128_1, conv128_2, conv256_down, conv256_1, conv256_2, conv256_down_out, fc; var weightMap, paramMappings, _a, extractConvLayerParams, extractResidualLayerParams, conv32_down, conv32_1, conv32_2, conv32_3, conv64_down, conv64_1, conv64_2, conv64_3, conv128_down, conv128_1, conv128_2, conv256_down, conv256_1, conv256_2, conv256_down_out, fc, params;
return tslib_1.__generator(this, function (_b) { return tslib_1.__generator(this, function (_b) {
switch (_b.label) { switch (_b.label) {
case 0: return [4 /*yield*/, loadWeightMap_1.loadWeightMap(uri, DEFAULT_MODEL_NAME)]; case 0: return [4 /*yield*/, loadWeightMap_1.loadWeightMap(uri, DEFAULT_MODEL_NAME)];
case 1: case 1:
weightMap = _b.sent(); weightMap = _b.sent();
_a = extractorsFactory(weightMap), extractConvLayerParams = _a.extractConvLayerParams, extractResidualLayerParams = _a.extractResidualLayerParams; paramMappings = [];
_a = extractorsFactory(weightMap, paramMappings), extractConvLayerParams = _a.extractConvLayerParams, extractResidualLayerParams = _a.extractResidualLayerParams;
conv32_down = extractConvLayerParams('conv32_down'); conv32_down = extractConvLayerParams('conv32_down');
conv32_1 = extractResidualLayerParams('conv32_1'); conv32_1 = extractResidualLayerParams('conv32_1');
conv32_2 = extractResidualLayerParams('conv32_2'); conv32_2 = extractResidualLayerParams('conv32_2');
...@@ -70,10 +56,11 @@ function loadQuantizedParams(uri) { ...@@ -70,10 +56,11 @@ function loadQuantizedParams(uri) {
conv256_2 = extractResidualLayerParams('conv256_2'); conv256_2 = extractResidualLayerParams('conv256_2');
conv256_down_out = extractResidualLayerParams('conv256_down_out'); conv256_down_out = extractResidualLayerParams('conv256_down_out');
fc = weightMap['fc']; fc = weightMap['fc'];
paramMappings.push({ originalPath: 'fc', paramPath: 'fc' });
if (!isTensor_1.isTensor2D(fc)) { if (!isTensor_1.isTensor2D(fc)) {
throw new Error("expected weightMap[fc] to be a Tensor2D, instead have " + fc); throw new Error("expected weightMap[fc] to be a Tensor2D, instead have " + fc);
} }
return [2 /*return*/, { params = {
conv32_down: conv32_down, conv32_down: conv32_down,
conv32_1: conv32_1, conv32_1: conv32_1,
conv32_2: conv32_2, conv32_2: conv32_2,
...@@ -90,7 +77,9 @@ function loadQuantizedParams(uri) { ...@@ -90,7 +77,9 @@ function loadQuantizedParams(uri) {
conv256_2: conv256_2, conv256_2: conv256_2,
conv256_down_out: conv256_down_out, conv256_down_out: conv256_down_out,
fc: fc fc: fc
}]; };
disposeUnusedWeightTensors_1.disposeUnusedWeightTensors(weightMap, paramMappings);
return [2 /*return*/, { params: params, paramMappings: paramMappings }];
} }
}); });
}); });
......
build/faceRecognitionNet/loadQuantizedParams.js.map
{"version":3,"file":"loadQuantizedParams.js","sourceRoot":"","sources":["../../src/faceRecognitionNet/loadQuantizedParams.ts"],"names":[],"mappings":";;;AAAA,gDAAyE;AACzE,0DAAyD;AAGzD,IAAM,kBAAkB,GAAG,wBAAwB,CAAA;AAEnD,2BAA2B,SAAc;IAEvC,iCAAiC,MAAc;QAC7C,IAAM,MAAM,GAAG;YACb,OAAO,EAAE,SAAS,CAAI,MAAM,mBAAgB,CAAC;YAC7C,MAAM,EAAE,SAAS,CAAI,MAAM,kBAAe,CAAC;SAC5C,CAAA;QAED,IAAI,CAAC,qBAAU,CAAC,MAAM,CAAC,OAAO,CAAC,EAAE;YAC/B,MAAM,IAAI,KAAK,CAAC,wBAAsB,MAAM,uDAAkD,MAAM,CAAC,OAAS,CAAC,CAAA;SAChH;QAED,IAAI,CAAC,qBAAU,CAAC,MAAM,CAAC,MAAM,CAAC,EAAE;YAC9B,MAAM,IAAI,KAAK,CAAC,wBAAsB,MAAM,sDAAiD,MAAM,CAAC,MAAQ,CAAC,CAAA;SAC9G;QAED,OAAO,MAAM,CAAA;IACf,CAAC;IAED,gCAAgC,MAAc;QAC5C,IAAM,MAAM,GAAG;YACb,OAAO,EAAE,SAAS,CAAI,MAAM,kBAAe,CAAC;YAC5C,IAAI,EAAE,SAAS,CAAI,MAAM,eAAY,CAAC;SACvC,CAAA;QAED,IAAI,CAAC,qBAAU,CAAC,MAAM,CAAC,OAAO,CAAC,EAAE;YAC/B,MAAM,IAAI,KAAK,CAAC,wBAAsB,MAAM,sDAAiD,MAAM,CAAC,OAAS,CAAC,CAAA;SAC/G;QAED,IAAI,CAAC,qBAAU,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE;YAC5B,MAAM,IAAI,KAAK,CAAC,wBAAsB,MAAM,mDAA8C,MAAM,CAAC,IAAM,CAAC,CAAA;SACzG;QAED,OAAO;YACL,IAAI,EAAE,MAAM;YACZ,KAAK,EAAE,uBAAuB,CAAC,MAAM,CAAC;SACvC,CAAA;IACH,CAAC;IAED,oCAAoC,MAAc;QAChD,OAAO;YACL,KAAK,EAAE,sBAAsB,CAAI,MAAM,WAAQ,CAAC;YAChD,KAAK,EAAE,sBAAsB,CAAI,MAAM,WAAQ,CAAC;SACjD,CAAA;IACH,CAAC;IAED,OAAO;QACL,sBAAsB,wBAAA;QACtB,0BAA0B,4BAAA;KAC3B,CAAA;AAEH,CAAC;AAED,6BAA0C,GAAuB;;;;;wBAC7C,qBAAM,6BAAa,CAAC,GAAG,EAAE,kBAAkB,CAAC,EAAA;;oBAAxD,SAAS,GAAG,SAA4C;oBAExD,KAGF,iBAAiB,CAAC,SAAS,CAAC,EAF9B,sBAAsB,4BAAA,EACtB,0BAA0B,gCAAA,CACI;oBAE1B,WAAW,GAAG,sBAAsB,CAAC,aAAa,CAAC,CAAA;oBACnD,QAAQ,GAAG,0BAA0B,CAAC,UAAU,CAAC,CAAA;oBACjD,QAAQ,GAAG,0BAA0B,CAAC,UAAU,CAAC,CAAA;oBACjD,QAAQ,GAAG,0BAA0B,CAAC,UAAU,CAAC,CAAA;oBAEjD,WAAW,GAAG,0BAA0B,CAAC,aAAa,CAAC,CAAA;oBACvD,QAAQ,GAAG,0BAA0B,CAAC,UAAU,CAAC,CAAA;oBACjD,QAAQ,GAAG,0BAA0B,CAAC,UAAU,CAAC,CAAA;oBACjD,QAAQ,GAAG,0BAA0B,CAAC,UAAU,CAAC,CAAA;oBAEjD,YAAY,GAAG,0BAA0B,CAAC,cAAc,CAAC,CAAA;oBACzD,SAAS,GAAG,0BAA0B,CAAC,WAAW,CAAC,CAAA;oBACnD,SAAS,GAAG,0BAA0B,CAAC,WAAW,CAAC,CAAA;oBAEnD,YAAY,GAAG,0BAA0B,CAAC,cAAc,CAAC,CAAA;oBACzD,SAAS,GAAG,0BAA0B,CAAC,WAAW,CAAC,CAAA;oBACnD,SAAS,GAAG,0BAA0B,CAAC,WAAW,CAAC,CAAA;oBACnD,gBAAgB,GAAG,0BAA0B,CAAC,kBAAkB,CAAC,CAAA;oBAEjE,EAAE,GAAG,SAAS,CAAC,IAAI,CAAC,CAAA;oBAE1B,IAAI,CAAC,qBAAU,CAAC,EAAE,CAAC,EAAE;wBACnB,MAAM,IAAI,KAAK,CAAC,2DAAyD,EAAI,CAAC,CAAA;qBAC/E;oBAED,sBAAO;4BACL,WAAW,aAAA;4BACX,QAAQ,UAAA;4BACR,QAAQ,UAAA;4BACR,QAAQ,UAAA;4BACR,WAAW,aAAA;4BACX,QAAQ,UAAA;4BACR,QAAQ,UAAA;4BACR,QAAQ,UAAA;4BACR,YAAY,cAAA;4BACZ,SAAS,WAAA;4BACT,SAAS,WAAA;4BACT,YAAY,cAAA;4BACZ,SAAS,WAAA;4BACT,SAAS,WAAA;4BACT,gBAAgB,kBAAA;4BAChB,EAAE,IAAA;yBACH,EAAA;;;;CACF;AAnDD,kDAmDC"} {"version":3,"file":"loadQuantizedParams.js","sourceRoot":"","sources":["../../src/faceRecognitionNet/loadQuantizedParams.ts"],"names":[],"mappings":";;;AAEA,oFAAmF;AACnF,kFAAiF;AACjF,gDAAiD;AACjD,0DAAyD;AAIzD,IAAM,kBAAkB,GAAG,wBAAwB,CAAA;AAEnD,2BAA2B,SAAc,EAAE,aAA6B;IAEtE,IAAM,kBAAkB,GAAG,qDAAyB,CAAC,SAAS,EAAE,aAAa,CAAC,CAAA;IAE9E,iCAAiC,MAAc;QAE7C,IAAM,OAAO,GAAG,kBAAkB,CAAiB,MAAM,mBAAgB,EAAE,CAAC,CAAC,CAAA;QAC7E,IAAM,MAAM,GAAG,kBAAkB,CAAiB,MAAM,kBAAe,EAAE,CAAC,CAAC,CAAA;QAE3E,OAAO,EAAE,OAAO,SAAA,EAAE,MAAM,QAAA,EAAE,CAAA;IAC5B,CAAC;IAED,gCAAgC,MAAc;QAE5C,IAAM,OAAO,GAAG,kBAAkB,CAAiB,MAAM,kBAAe,EAAE,CAAC,CAAC,CAAA;QAC5E,IAAM,IAAI,GAAG,kBAAkB,CAAiB,MAAM,eAAY,EAAE,CAAC,CAAC,CAAA;QACtE,IAAM,KAAK,GAAG,uBAAuB,CAAC,MAAM,CAAC,CAAA;QAE7C,OAAO,EAAE,IAAI,EAAE,EAAE,OAAO,SAAA,EAAE,IAAI,MAAA,EAAE,EAAE,KAAK,OAAA,EAAE,CAAA;IAC3C,CAAC;IAED,oCAAoC,MAAc;QAChD,OAAO;YACL,KAAK,EAAE,sBAAsB,CAAI,MAAM,WAAQ,CAAC;YAChD,KAAK,EAAE,sBAAsB,CAAI,MAAM,WAAQ,CAAC;SACjD,CAAA;IACH,CAAC;IAED,OAAO;QACL,sBAAsB,wBAAA;QACtB,0BAA0B,4BAAA;KAC3B,CAAA;AAEH,CAAC;AAED,6BACE,GAAuB;;;;;wBAGL,qBAAM,6BAAa,CAAC,GAAG,EAAE,kBAAkB,CAAC,EAAA;;oBAAxD,SAAS,GAAG,SAA4C;oBACxD,aAAa,GAAmB,EAAE,CAAA;oBAElC,KAGF,iBAAiB,CAAC,SAAS,EAAE,aAAa,CAAC,EAF7C,sBAAsB,4BAAA,EACtB,0BAA0B,gCAAA,CACmB;oBAEzC,WAAW,GAAG,sBAAsB,CAAC,aAAa,CAAC,CAAA;oBACnD,QAAQ,GAAG,0BAA0B,CAAC,UAAU,CAAC,CAAA;oBACjD,QAAQ,GAAG,0BAA0B,CAAC,UAAU,CAAC,CAAA;oBACjD,QAAQ,GAAG,0BAA0B,CAAC,UAAU,CAAC,CAAA;oBAEjD,WAAW,GAAG,0BAA0B,CAAC,aAAa,CAAC,CAAA;oBACvD,QAAQ,GAAG,0BAA0B,CAAC,UAAU,CAAC,CAAA;oBACjD,QAAQ,GAAG,0BAA0B,CAAC,UAAU,CAAC,CAAA;oBACjD,QAAQ,GAAG,0BAA0B,CAAC,UAAU,CAAC,CAAA;oBAEjD,YAAY,GAAG,0BAA0B,CAAC,cAAc,CAAC,CAAA;oBACzD,SAAS,GAAG,0BAA0B,CAAC,WAAW,CAAC,CAAA;oBACnD,SAAS,GAAG,0BAA0B,CAAC,WAAW,CAAC,CAAA;oBAEnD,YAAY,GAAG,0BAA0B,CAAC,cAAc,CAAC,CAAA;oBACzD,SAAS,GAAG,0BAA0B,CAAC,WAAW,CAAC,CAAA;oBACnD,SAAS,GAAG,0BAA0B,CAAC,WAAW,CAAC,CAAA;oBACnD,gBAAgB,GAAG,0BAA0B,CAAC,kBAAkB,CAAC,CAAA;oBAEjE,EAAE,GAAG,SAAS,CAAC,IAAI,CAAC,CAAA;oBAC1B,aAAa,CAAC,IAAI,CAAC,EAAE,YAAY,EAAE,IAAI,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;oBAE3D,IAAI,CAAC,qBAAU,CAAC,EAAE,CAAC,EAAE;wBACnB,MAAM,IAAI,KAAK,CAAC,2DAAyD,EAAI,CAAC,CAAA;qBAC/E;oBAEK,MAAM,GAAG;wBACb,WAAW,aAAA;wBACX,QAAQ,UAAA;wBACR,QAAQ,UAAA;wBACR,QAAQ,UAAA;wBACR,WAAW,aAAA;wBACX,QAAQ,UAAA;wBACR,QAAQ,UAAA;wBACR,QAAQ,UAAA;wBACR,YAAY,cAAA;wBACZ,SAAS,WAAA;wBACT,SAAS,WAAA;wBACT,YAAY,cAAA;wBACZ,SAAS,WAAA;wBACT,SAAS,WAAA;wBACT,gBAAgB,kBAAA;wBAChB,EAAE,IAAA;qBACH,CAAA;oBAED,uDAA0B,CAAC,SAAS,EAAE,aAAa,CAAC,CAAA;oBAEpD,sBAAO,EAAE,MAAM,QAAA,EAAE,aAAa,eAAA,EAAE,EAAA;;;;CACjC;AA5DD,kDA4DC"}
\ No newline at end of file \ No newline at end of file
dist/face-api.js
...@@ -271,9 +271,6 @@ ...@@ -271,9 +271,6 @@
function isTensor(tensor$$1, dim) { function isTensor(tensor$$1, dim) {
return tensor$$1 instanceof Tensor && tensor$$1.shape.length === dim; return tensor$$1 instanceof Tensor && tensor$$1.shape.length === dim;
} }
function isTensor1D(tensor$$1) {
return isTensor(tensor$$1, 1);
}
function isTensor2D(tensor$$1) { function isTensor2D(tensor$$1) {
return isTensor(tensor$$1, 2); return isTensor(tensor$$1, 2);
} }
...@@ -663,6 +660,10 @@ ...@@ -663,6 +660,10 @@
} }
return new Rect(x, y, width, height); return new Rect(x, y, width, height);
}; };
Rect.prototype.pad = function (padX, padY) {
var _a = this, x = _a.x, y = _a.y, width = _a.width, height = _a.height;
return new Rect(x - (padX / 2), y - (padY / 2), width + padX, height + padY);
};
Rect.prototype.floor = function () { Rect.prototype.floor = function () {
return new Rect(Math.floor(this.x), Math.floor(this.y), Math.floor(this.width), Math.floor(this.height)); return new Rect(Math.floor(this.x), Math.floor(this.y), Math.floor(this.width), Math.floor(this.height));
}; };
...@@ -955,6 +956,128 @@ ...@@ -955,6 +956,128 @@
}); });
} }
var NeuralNetwork = /** @class */ (function () {
function NeuralNetwork(_name) {
this._name = _name;
this._params = undefined;
this._paramMappings = [];
}
Object.defineProperty(NeuralNetwork.prototype, "params", {
get: function () {
return this._params;
},
enumerable: true,
configurable: true
});
Object.defineProperty(NeuralNetwork.prototype, "paramMappings", {
get: function () {
return this._paramMappings;
},
enumerable: true,
configurable: true
});
NeuralNetwork.prototype.getParamFromPath = function (paramPath) {
var _a = this.traversePropertyPath(paramPath), obj = _a.obj, objProp = _a.objProp;
return obj[objProp];
};
NeuralNetwork.prototype.reassignParamFromPath = function (paramPath, tensor$$1) {
var _a = this.traversePropertyPath(paramPath), obj = _a.obj, objProp = _a.objProp;
obj[objProp].dispose();
obj[objProp] = tensor$$1;
};
NeuralNetwork.prototype.getParamList = function () {
var _this = this;
return this._paramMappings.map(function (_a) {
var paramPath = _a.paramPath;
return ({
path: paramPath,
tensor: _this.getParamFromPath(paramPath)
});
});
};
NeuralNetwork.prototype.getTrainableParams = function () {
return this.getParamList().filter(function (param) { return param.tensor instanceof Variable; });
};
NeuralNetwork.prototype.getFrozenParams = function () {
return this.getParamList().filter(function (param) { return !(param.tensor instanceof Variable); });
};
NeuralNetwork.prototype.variable = function () {
var _this = this;
this.getFrozenParams().forEach(function (_a) {
var path = _a.path, tensor$$1 = _a.tensor;
_this.reassignParamFromPath(path, variable(tensor$$1));
});
};
NeuralNetwork.prototype.freeze = function () {
var _this = this;
this.getTrainableParams().forEach(function (_a) {
var path = _a.path, tensor$$1 = _a.tensor;
_this.reassignParamFromPath(path, tensor(tensor$$1));
});
};
NeuralNetwork.prototype.dispose = function (throwOnRedispose) {
if (throwOnRedispose === void 0) { throwOnRedispose = true; }
this.getParamList().forEach(function (param) {
if (throwOnRedispose && param.tensor.isDisposed) {
throw new Error("param tensor has already been disposed for path " + param.path);
}
param.tensor.dispose();
});
this._params = undefined;
};
NeuralNetwork.prototype.load = function (weightsOrUrl) {
return __awaiter$1(this, void 0, void 0, function () {
var _a, paramMappings, params;
return __generator$1(this, function (_b) {
switch (_b.label) {
case 0:
if (weightsOrUrl instanceof Float32Array) {
this.extractWeights(weightsOrUrl);
return [2 /*return*/];
}
if (weightsOrUrl && typeof weightsOrUrl !== 'string') {
throw new Error(this._name + ".load - expected model uri, or weights as Float32Array");
}
return [4 /*yield*/, this.loadQuantizedParams(weightsOrUrl)];
case 1:
_a = _b.sent(), paramMappings = _a.paramMappings, params = _a.params;
this._paramMappings = paramMappings;
this._params = params;
return [2 /*return*/];
}
});
});
};
NeuralNetwork.prototype.extractWeights = function (weights) {
var _a = this.extractParams(weights), paramMappings = _a.paramMappings, params = _a.params;
this._paramMappings = paramMappings;
this._params = params;
};
NeuralNetwork.prototype.traversePropertyPath = function (paramPath) {
if (!this.params) {
throw new Error("traversePropertyPath - model has no loaded params");
}
var result = paramPath.split('/').reduce(function (res, objProp) {
if (!res.nextObj.hasOwnProperty(objProp)) {
throw new Error("traversePropertyPath - object does not have property " + objProp + ", for path " + paramPath);
}
return { obj: res.nextObj, objProp: objProp, nextObj: res.nextObj[objProp] };
}, { nextObj: this.params });
var obj = result.obj, objProp = result.objProp;
if (!obj || !objProp || !(obj[objProp] instanceof Tensor)) {
throw new Error("traversePropertyPath - parameter is not a tensor, for path " + paramPath);
}
return { obj: obj, objProp: objProp };
};
NeuralNetwork.prototype.loadQuantizedParams = function (_) {
throw new Error(this._name + ".loadQuantizedParams - not implemented");
};
NeuralNetwork.prototype.extractParams = function (_) {
throw new Error(this._name + ".extractParams - not implemented");
};
return NeuralNetwork;
}());
function extractWeightsFactory(weights) { function extractWeightsFactory(weights) {
var remainingWeights = weights; var remainingWeights = weights;
function extractWeights(numWeights) { function extractWeights(numWeights) {
...@@ -971,13 +1094,14 @@ ...@@ -971,13 +1094,14 @@
}; };
} }
function extractorsFactory(extractWeights) { function extractorsFactory(extractWeights, paramMappings) {
function extractDepthwiseConvParams(numChannels) { function extractDepthwiseConvParams(numChannels, mappedPrefix) {
var filters = tensor4d(extractWeights(3 * 3 * numChannels), [3, 3, numChannels, 1]); var filters = tensor4d(extractWeights(3 * 3 * numChannels), [3, 3, numChannels, 1]);
var batch_norm_scale = tensor1d(extractWeights(numChannels)); var batch_norm_scale = tensor1d(extractWeights(numChannels));
var batch_norm_offset = tensor1d(extractWeights(numChannels)); var batch_norm_offset = tensor1d(extractWeights(numChannels));
var batch_norm_mean = tensor1d(extractWeights(numChannels)); var batch_norm_mean = tensor1d(extractWeights(numChannels));
var batch_norm_variance = tensor1d(extractWeights(numChannels)); var batch_norm_variance = tensor1d(extractWeights(numChannels));
paramMappings.push({ paramPath: mappedPrefix + "/filters" }, { paramPath: mappedPrefix + "/batch_norm_scale" }, { paramPath: mappedPrefix + "/batch_norm_offset" }, { paramPath: mappedPrefix + "/batch_norm_mean" }, { paramPath: mappedPrefix + "/batch_norm_variance" });
return { return {
filters: filters, filters: filters,
batch_norm_scale: batch_norm_scale, batch_norm_scale: batch_norm_scale,
...@@ -986,115 +1110,116 @@ ...@@ -986,115 +1110,116 @@
batch_norm_variance: batch_norm_variance batch_norm_variance: batch_norm_variance
}; };
} }
function extractConvParams(channelsIn, channelsOut, filterSize) { function extractConvParams(channelsIn, channelsOut, filterSize, mappedPrefix, isPointwiseConv) {
var filters = tensor4d(extractWeights(channelsIn * channelsOut * filterSize * filterSize), [filterSize, filterSize, channelsIn, channelsOut]); var filters = tensor4d(extractWeights(channelsIn * channelsOut * filterSize * filterSize), [filterSize, filterSize, channelsIn, channelsOut]);
var bias = tensor1d(extractWeights(channelsOut)); var bias = tensor1d(extractWeights(channelsOut));
return { paramMappings.push({ paramPath: mappedPrefix + "/filters" }, { paramPath: mappedPrefix + "/" + (isPointwiseConv ? 'batch_norm_offset' : 'bias') });
filters: filters, return { filters: filters, bias: bias };
bias: bias
};
} }
function extractPointwiseConvParams(channelsIn, channelsOut, filterSize) { function extractPointwiseConvParams(channelsIn, channelsOut, filterSize, mappedPrefix) {
var _a = extractConvParams(channelsIn, channelsOut, filterSize), filters = _a.filters, bias = _a.bias; var _a = extractConvParams(channelsIn, channelsOut, filterSize, mappedPrefix, true), filters = _a.filters, bias = _a.bias;
return { return {
filters: filters, filters: filters,
batch_norm_offset: bias batch_norm_offset: bias
}; };
} }
function extractConvPairParams(channelsIn, channelsOut) { function extractConvPairParams(channelsIn, channelsOut, mappedPrefix) {
var depthwise_conv_params = extractDepthwiseConvParams(channelsIn); var depthwise_conv = extractDepthwiseConvParams(channelsIn, mappedPrefix + "/depthwise_conv");
var pointwise_conv_params = extractPointwiseConvParams(channelsIn, channelsOut, 1); var pointwise_conv = extractPointwiseConvParams(channelsIn, channelsOut, 1, mappedPrefix + "/pointwise_conv");
return { return { depthwise_conv: depthwise_conv, pointwise_conv: pointwise_conv };
depthwise_conv_params: depthwise_conv_params,
pointwise_conv_params: pointwise_conv_params
};
} }
function extractMobilenetV1Params() { function extractMobilenetV1Params() {
var conv_0_params = extractPointwiseConvParams(3, 32, 3); var conv_0 = extractPointwiseConvParams(3, 32, 3, 'mobilenetv1/conv_0');
var channelNumPairs = [ var conv_1 = extractConvPairParams(32, 64, 'mobilenetv1/conv_1');
[32, 64], var conv_2 = extractConvPairParams(64, 128, 'mobilenetv1/conv_2');
[64, 128], var conv_3 = extractConvPairParams(128, 128, 'mobilenetv1/conv_3');
[128, 128], var conv_4 = extractConvPairParams(128, 256, 'mobilenetv1/conv_4');
[128, 256], var conv_5 = extractConvPairParams(256, 256, 'mobilenetv1/conv_5');
[256, 256], var conv_6 = extractConvPairParams(256, 512, 'mobilenetv1/conv_6');
[256, 512], var conv_7 = extractConvPairParams(512, 512, 'mobilenetv1/conv_7');
[512, 512], var conv_8 = extractConvPairParams(512, 512, 'mobilenetv1/conv_8');
[512, 512], var conv_9 = extractConvPairParams(512, 512, 'mobilenetv1/conv_9');
[512, 512], var conv_10 = extractConvPairParams(512, 512, 'mobilenetv1/conv_10');
[512, 512], var conv_11 = extractConvPairParams(512, 512, 'mobilenetv1/conv_11');
[512, 512], var conv_12 = extractConvPairParams(512, 1024, 'mobilenetv1/conv_12');
[512, 1024], var conv_13 = extractConvPairParams(1024, 1024, 'mobilenetv1/conv_13');
[1024, 1024]
];
var conv_pair_params = channelNumPairs.map(function (_a) {
var channelsIn = _a[0], channelsOut = _a[1];
return extractConvPairParams(channelsIn, channelsOut);
});
return { return {
conv_0_params: conv_0_params, conv_0: conv_0,
conv_pair_params: conv_pair_params conv_1: conv_1,
conv_2: conv_2,
conv_3: conv_3,
conv_4: conv_4,
conv_5: conv_5,
conv_6: conv_6,
conv_7: conv_7,
conv_8: conv_8,
conv_9: conv_9,
conv_10: conv_10,
conv_11: conv_11,
conv_12: conv_12,
conv_13: conv_13
}; };
} }
function extractPredictionLayerParams() { function extractPredictionLayerParams() {
var conv_0_params = extractPointwiseConvParams(1024, 256, 1); var conv_0 = extractPointwiseConvParams(1024, 256, 1, 'prediction_layer/conv_0');
var conv_1_params = extractPointwiseConvParams(256, 512, 3); var conv_1 = extractPointwiseConvParams(256, 512, 3, 'prediction_layer/conv_1');
var conv_2_params = extractPointwiseConvParams(512, 128, 1); var conv_2 = extractPointwiseConvParams(512, 128, 1, 'prediction_layer/conv_2');
var conv_3_params = extractPointwiseConvParams(128, 256, 3); var conv_3 = extractPointwiseConvParams(128, 256, 3, 'prediction_layer/conv_3');
var conv_4_params = extractPointwiseConvParams(256, 128, 1); var conv_4 = extractPointwiseConvParams(256, 128, 1, 'prediction_layer/conv_4');
var conv_5_params = extractPointwiseConvParams(128, 256, 3); var conv_5 = extractPointwiseConvParams(128, 256, 3, 'prediction_layer/conv_5');
var conv_6_params = extractPointwiseConvParams(256, 64, 1); var conv_6 = extractPointwiseConvParams(256, 64, 1, 'prediction_layer/conv_6');
var conv_7_params = extractPointwiseConvParams(64, 128, 3); var conv_7 = extractPointwiseConvParams(64, 128, 3, 'prediction_layer/conv_7');
var box_encoding_0_predictor_params = extractConvParams(512, 12, 1); var box_encoding_0_predictor = extractConvParams(512, 12, 1, 'prediction_layer/box_predictor_0/box_encoding_predictor');
var class_predictor_0_params = extractConvParams(512, 9, 1); var class_predictor_0 = extractConvParams(512, 9, 1, 'prediction_layer/box_predictor_0/class_predictor');
var box_encoding_1_predictor_params = extractConvParams(1024, 24, 1); var box_encoding_1_predictor = extractConvParams(1024, 24, 1, 'prediction_layer/box_predictor_1/box_encoding_predictor');
var class_predictor_1_params = extractConvParams(1024, 18, 1); var class_predictor_1 = extractConvParams(1024, 18, 1, 'prediction_layer/box_predictor_1/class_predictor');
var box_encoding_2_predictor_params = extractConvParams(512, 24, 1); var box_encoding_2_predictor = extractConvParams(512, 24, 1, 'prediction_layer/box_predictor_2/box_encoding_predictor');
var class_predictor_2_params = extractConvParams(512, 18, 1); var class_predictor_2 = extractConvParams(512, 18, 1, 'prediction_layer/box_predictor_2/class_predictor');
var box_encoding_3_predictor_params = extractConvParams(256, 24, 1); var box_encoding_3_predictor = extractConvParams(256, 24, 1, 'prediction_layer/box_predictor_3/box_encoding_predictor');
var class_predictor_3_params = extractConvParams(256, 18, 1); var class_predictor_3 = extractConvParams(256, 18, 1, 'prediction_layer/box_predictor_3/class_predictor');
var box_encoding_4_predictor_params = extractConvParams(256, 24, 1); var box_encoding_4_predictor = extractConvParams(256, 24, 1, 'prediction_layer/box_predictor_4/box_encoding_predictor');
var class_predictor_4_params = extractConvParams(256, 18, 1); var class_predictor_4 = extractConvParams(256, 18, 1, 'prediction_layer/box_predictor_4/class_predictor');
var box_encoding_5_predictor_params = extractConvParams(128, 24, 1); var box_encoding_5_predictor = extractConvParams(128, 24, 1, 'prediction_layer/box_predictor_5/box_encoding_predictor');
var class_predictor_5_params = extractConvParams(128, 18, 1); var class_predictor_5 = extractConvParams(128, 18, 1, 'prediction_layer/box_predictor_5/class_predictor');
var box_predictor_0_params = { var box_predictor_0 = {
box_encoding_predictor_params: box_encoding_0_predictor_params, box_encoding_predictor: box_encoding_0_predictor,
class_predictor_params: class_predictor_0_params class_predictor: class_predictor_0
}; };
var box_predictor_1_params = { var box_predictor_1 = {
box_encoding_predictor_params: box_encoding_1_predictor_params, box_encoding_predictor: box_encoding_1_predictor,
class_predictor_params: class_predictor_1_params class_predictor: class_predictor_1
}; };
var box_predictor_2_params = { var box_predictor_2 = {
box_encoding_predictor_params: box_encoding_2_predictor_params, box_encoding_predictor: box_encoding_2_predictor,
class_predictor_params: class_predictor_2_params class_predictor: class_predictor_2
}; };
var box_predictor_3_params = { var box_predictor_3 = {
box_encoding_predictor_params: box_encoding_3_predictor_params, box_encoding_predictor: box_encoding_3_predictor,
class_predictor_params: class_predictor_3_params class_predictor: class_predictor_3
}; };
var box_predictor_4_params = { var box_predictor_4 = {
box_encoding_predictor_params: box_encoding_4_predictor_params, box_encoding_predictor: box_encoding_4_predictor,
class_predictor_params: class_predictor_4_params class_predictor: class_predictor_4
}; };
var box_predictor_5_params = { var box_predictor_5 = {
box_encoding_predictor_params: box_encoding_5_predictor_params, box_encoding_predictor: box_encoding_5_predictor,
class_predictor_params: class_predictor_5_params class_predictor: class_predictor_5
}; };
return { return {
conv_0_params: conv_0_params, conv_0: conv_0,
conv_1_params: conv_1_params, conv_1: conv_1,
conv_2_params: conv_2_params, conv_2: conv_2,
conv_3_params: conv_3_params, conv_3: conv_3,
conv_4_params: conv_4_params, conv_4: conv_4,
conv_5_params: conv_5_params, conv_5: conv_5,
conv_6_params: conv_6_params, conv_6: conv_6,
conv_7_params: conv_7_params, conv_7: conv_7,
box_predictor_0_params: box_predictor_0_params, box_predictor_0: box_predictor_0,
box_predictor_1_params: box_predictor_1_params, box_predictor_1: box_predictor_1,
box_predictor_2_params: box_predictor_2_params, box_predictor_2: box_predictor_2,
box_predictor_3_params: box_predictor_3_params, box_predictor_3: box_predictor_3,
box_predictor_4_params: box_predictor_4_params, box_predictor_4: box_predictor_4,
box_predictor_5_params: box_predictor_5_params box_predictor_5: box_predictor_5
}; };
} }
return { return {
...@@ -1103,21 +1228,45 @@ ...@@ -1103,21 +1228,45 @@
}; };
} }
function extractParams(weights) { function extractParams(weights) {
var paramMappings = [];
var _a = extractWeightsFactory(weights), extractWeights = _a.extractWeights, getRemainingWeights = _a.getRemainingWeights; var _a = extractWeightsFactory(weights), extractWeights = _a.extractWeights, getRemainingWeights = _a.getRemainingWeights;
var _b = extractorsFactory(extractWeights), extractMobilenetV1Params = _b.extractMobilenetV1Params, extractPredictionLayerParams = _b.extractPredictionLayerParams; var _b = extractorsFactory(extractWeights, paramMappings), extractMobilenetV1Params = _b.extractMobilenetV1Params, extractPredictionLayerParams = _b.extractPredictionLayerParams;
var mobilenetv1_params = extractMobilenetV1Params(); var mobilenetv1 = extractMobilenetV1Params();
var prediction_layer_params = extractPredictionLayerParams(); var prediction_layer = extractPredictionLayerParams();
var extra_dim = tensor3d(extractWeights(5118 * 4), [1, 5118, 4]); var extra_dim = tensor3d(extractWeights(5118 * 4), [1, 5118, 4]);
var output_layer_params = { var output_layer = {
extra_dim: extra_dim extra_dim: extra_dim
}; };
paramMappings.push({ paramPath: 'output_layer/extra_dim' });
if (getRemainingWeights().length !== 0) { if (getRemainingWeights().length !== 0) {
throw new Error("weights remaing after extract: " + getRemainingWeights().length); throw new Error("weights remaing after extract: " + getRemainingWeights().length);
} }
return { return {
mobilenetv1_params: mobilenetv1_params, params: {
prediction_layer_params: prediction_layer_params, mobilenetv1: mobilenetv1,
output_layer_params: output_layer_params prediction_layer: prediction_layer,
output_layer: output_layer
},
paramMappings: paramMappings
};
}
function disposeUnusedWeightTensors(weightMap, paramMappings) {
Object.keys(weightMap).forEach(function (path) {
if (!paramMappings.some(function (pm) { return pm.originalPath === path; })) {
weightMap[path].dispose();
}
});
}
function extractWeightEntryFactory(weightMap, paramMappings) {
return function (originalPath, paramRank, mappedPath) {
var tensor = weightMap[originalPath];
if (!isTensor(tensor, paramRank)) {
throw new Error("expected weightMap[" + originalPath + "] to be a Tensor" + paramRank + "D, instead have " + tensor);
}
paramMappings.push({ originalPath: originalPath, paramPath: mappedPath || originalPath });
return tensor;
}; };
} }
...@@ -1152,95 +1301,78 @@ ...@@ -1152,95 +1301,78 @@
} }
var DEFAULT_MODEL_NAME = 'face_detection_model'; var DEFAULT_MODEL_NAME = 'face_detection_model';
function extractorsFactory$1(weightMap) { function extractorsFactory$1(weightMap, paramMappings) {
function extractPointwiseConvParams(prefix, idx) { var extractWeightEntry = extractWeightEntryFactory(weightMap, paramMappings);
var pointwise_conv_params = { function extractPointwiseConvParams(prefix, idx, mappedPrefix) {
filters: weightMap[prefix + "/Conv2d_" + idx + "_pointwise/weights"], var filters = extractWeightEntry(prefix + "/Conv2d_" + idx + "_pointwise/weights", 4, mappedPrefix + "/filters");
batch_norm_offset: weightMap[prefix + "/Conv2d_" + idx + "_pointwise/convolution_bn_offset"] var batch_norm_offset = extractWeightEntry(prefix + "/Conv2d_" + idx + "_pointwise/convolution_bn_offset", 1, mappedPrefix + "/batch_norm_offset");
}; return { filters: filters, batch_norm_offset: batch_norm_offset };
if (!isTensor4D(pointwise_conv_params.filters)) {
throw new Error("expected weightMap[" + prefix + "/Conv2d_" + idx + "_pointwise/weights] to be a Tensor4D, instead have " + pointwise_conv_params.filters);
}
if (!isTensor1D(pointwise_conv_params.batch_norm_offset)) {
throw new Error("expected weightMap[" + prefix + "/Conv2d_" + idx + "_pointwise/convolution_bn_offset] to be a Tensor1D, instead have " + pointwise_conv_params.batch_norm_offset);
}
return pointwise_conv_params;
} }
function extractConvPairParams(idx) { function extractConvPairParams(idx) {
var depthwise_conv_params = { var mappedPrefix = "mobilenetv1/conv_" + idx;
filters: weightMap["MobilenetV1/Conv2d_" + idx + "_depthwise/depthwise_weights"], var prefixDepthwiseConv = "MobilenetV1/Conv2d_" + idx + "_depthwise";
batch_norm_scale: weightMap["MobilenetV1/Conv2d_" + idx + "_depthwise/BatchNorm/gamma"], var mappedPrefixDepthwiseConv = mappedPrefix + "/depthwise_conv";
batch_norm_offset: weightMap["MobilenetV1/Conv2d_" + idx + "_depthwise/BatchNorm/beta"], var mappedPrefixPointwiseConv = mappedPrefix + "/pointwise_conv";
batch_norm_mean: weightMap["MobilenetV1/Conv2d_" + idx + "_depthwise/BatchNorm/moving_mean"], var filters = extractWeightEntry(prefixDepthwiseConv + "/depthwise_weights", 4, mappedPrefixDepthwiseConv + "/filters");
batch_norm_variance: weightMap["MobilenetV1/Conv2d_" + idx + "_depthwise/BatchNorm/moving_variance"], var batch_norm_scale = extractWeightEntry(prefixDepthwiseConv + "/BatchNorm/gamma", 1, mappedPrefixDepthwiseConv + "/batch_norm_scale");
}; var batch_norm_offset = extractWeightEntry(prefixDepthwiseConv + "/BatchNorm/beta", 1, mappedPrefixDepthwiseConv + "/batch_norm_offset");
if (!isTensor4D(depthwise_conv_params.filters)) { var batch_norm_mean = extractWeightEntry(prefixDepthwiseConv + "/BatchNorm/moving_mean", 1, mappedPrefixDepthwiseConv + "/batch_norm_mean");
throw new Error("expected weightMap[MobilenetV1/Conv2d_" + idx + "_depthwise/depthwise_weights] to be a Tensor4D, instead have " + depthwise_conv_params.filters); var batch_norm_variance = extractWeightEntry(prefixDepthwiseConv + "/BatchNorm/moving_variance", 1, mappedPrefixDepthwiseConv + "/batch_norm_variance");
}
if (!isTensor1D(depthwise_conv_params.batch_norm_scale)) {
throw new Error("expected weightMap[MobilenetV1/Conv2d_" + idx + "_depthwise/BatchNorm/gamma] to be a Tensor1D, instead have " + depthwise_conv_params.batch_norm_scale);
}
if (!isTensor1D(depthwise_conv_params.batch_norm_offset)) {
throw new Error("expected weightMap[MobilenetV1/Conv2d_" + idx + "_depthwise/BatchNorm/beta] to be a Tensor1D, instead have " + depthwise_conv_params.batch_norm_offset);
}
if (!isTensor1D(depthwise_conv_params.batch_norm_mean)) {
throw new Error("expected weightMap[MobilenetV1/Conv2d_" + idx + "_depthwise/BatchNorm/moving_mean] to be a Tensor1D, instead have " + depthwise_conv_params.batch_norm_mean);
}
if (!isTensor1D(depthwise_conv_params.batch_norm_variance)) {
throw new Error("expected weightMap[MobilenetV1/Conv2d_" + idx + "_depthwise/BatchNorm/moving_variance] to be a Tensor1D, instead have " + depthwise_conv_params.batch_norm_variance);
}
return { return {
depthwise_conv_params: depthwise_conv_params, depthwise_conv: {
pointwise_conv_params: extractPointwiseConvParams('MobilenetV1', idx) filters: filters,
batch_norm_scale: batch_norm_scale,
batch_norm_offset: batch_norm_offset,
batch_norm_mean: batch_norm_mean,
batch_norm_variance: batch_norm_variance
},
pointwise_conv: extractPointwiseConvParams('MobilenetV1', idx, mappedPrefixPointwiseConv)
}; };
} }
function extractMobilenetV1Params() { function extractMobilenetV1Params() {
return { return {
conv_0_params: extractPointwiseConvParams('MobilenetV1', 0), conv_0: extractPointwiseConvParams('MobilenetV1', 0, 'mobilenetv1/conv_0'),
conv_pair_params: Array(13).fill(0).map(function (_, i) { return extractConvPairParams(i + 1); }) conv_1: extractConvPairParams(1),
}; conv_2: extractConvPairParams(2),
} conv_3: extractConvPairParams(3),
function extractBoxPredictorParams(idx) { conv_4: extractConvPairParams(4),
var params = { conv_5: extractConvPairParams(5),
box_encoding_predictor_params: { conv_6: extractConvPairParams(6),
filters: weightMap["Prediction/BoxPredictor_" + idx + "/BoxEncodingPredictor/weights"], conv_7: extractConvPairParams(7),
bias: weightMap["Prediction/BoxPredictor_" + idx + "/BoxEncodingPredictor/biases"] conv_8: extractConvPairParams(8),
}, conv_9: extractConvPairParams(9),
class_predictor_params: { conv_10: extractConvPairParams(10),
filters: weightMap["Prediction/BoxPredictor_" + idx + "/ClassPredictor/weights"], conv_11: extractConvPairParams(11),
bias: weightMap["Prediction/BoxPredictor_" + idx + "/ClassPredictor/biases"] conv_12: extractConvPairParams(12),
} conv_13: extractConvPairParams(13)
}; };
if (!isTensor4D(params.box_encoding_predictor_params.filters)) {
throw new Error("expected weightMap[Prediction/BoxPredictor_" + idx + "/BoxEncodingPredictor/weights] to be a Tensor4D, instead have " + params.box_encoding_predictor_params.filters);
}
if (!isTensor1D(params.box_encoding_predictor_params.bias)) {
throw new Error("expected weightMap[Prediction/BoxPredictor_" + idx + "/BoxEncodingPredictor/biases] to be a Tensor1D, instead have " + params.box_encoding_predictor_params.bias);
}
if (!isTensor4D(params.class_predictor_params.filters)) {
throw new Error("expected weightMap[Prediction/BoxPredictor_" + idx + "/ClassPredictor/weights] to be a Tensor4D, instead have " + params.class_predictor_params.filters);
} }
if (!isTensor1D(params.class_predictor_params.bias)) { function extractConvParams(prefix, mappedPrefix) {
throw new Error("expected weightMap[Prediction/BoxPredictor_" + idx + "/ClassPredictor/biases] to be a Tensor1D, instead have " + params.class_predictor_params.bias); var filters = extractWeightEntry(prefix + "/weights", 4, mappedPrefix + "/filters");
var bias = extractWeightEntry(prefix + "/biases", 1, mappedPrefix + "/bias");
return { filters: filters, bias: bias };
} }
return params; function extractBoxPredictorParams(idx) {
var box_encoding_predictor = extractConvParams("Prediction/BoxPredictor_" + idx + "/BoxEncodingPredictor", "prediction_layer/box_predictor_" + idx + "/box_encoding_predictor");
var class_predictor = extractConvParams("Prediction/BoxPredictor_" + idx + "/ClassPredictor", "prediction_layer/box_predictor_" + idx + "/class_predictor");
return { box_encoding_predictor: box_encoding_predictor, class_predictor: class_predictor };
} }
function extractPredictionLayerParams() { function extractPredictionLayerParams() {
return { return {
conv_0_params: extractPointwiseConvParams('Prediction', 0), conv_0: extractPointwiseConvParams('Prediction', 0, 'prediction_layer/conv_0'),
conv_1_params: extractPointwiseConvParams('Prediction', 1), conv_1: extractPointwiseConvParams('Prediction', 1, 'prediction_layer/conv_1'),
conv_2_params: extractPointwiseConvParams('Prediction', 2), conv_2: extractPointwiseConvParams('Prediction', 2, 'prediction_layer/conv_2'),
conv_3_params: extractPointwiseConvParams('Prediction', 3), conv_3: extractPointwiseConvParams('Prediction', 3, 'prediction_layer/conv_3'),
conv_4_params: extractPointwiseConvParams('Prediction', 4), conv_4: extractPointwiseConvParams('Prediction', 4, 'prediction_layer/conv_4'),
conv_5_params: extractPointwiseConvParams('Prediction', 5), conv_5: extractPointwiseConvParams('Prediction', 5, 'prediction_layer/conv_5'),
conv_6_params: extractPointwiseConvParams('Prediction', 6), conv_6: extractPointwiseConvParams('Prediction', 6, 'prediction_layer/conv_6'),
conv_7_params: extractPointwiseConvParams('Prediction', 7), conv_7: extractPointwiseConvParams('Prediction', 7, 'prediction_layer/conv_7'),
box_predictor_0_params: extractBoxPredictorParams(0), box_predictor_0: extractBoxPredictorParams(0),
box_predictor_1_params: extractBoxPredictorParams(1), box_predictor_1: extractBoxPredictorParams(1),
box_predictor_2_params: extractBoxPredictorParams(2), box_predictor_2: extractBoxPredictorParams(2),
box_predictor_3_params: extractBoxPredictorParams(3), box_predictor_3: extractBoxPredictorParams(3),
box_predictor_4_params: extractBoxPredictorParams(4), box_predictor_4: extractBoxPredictorParams(4),
box_predictor_5_params: extractBoxPredictorParams(5) box_predictor_5: extractBoxPredictorParams(5)
}; };
} }
return { return {
...@@ -1250,24 +1382,28 @@ ...@@ -1250,24 +1382,28 @@
} }
function loadQuantizedParams(uri) { function loadQuantizedParams(uri) {
return __awaiter$1(this, void 0, void 0, function () { return __awaiter$1(this, void 0, void 0, function () {
var weightMap, _a, extractMobilenetV1Params, extractPredictionLayerParams, extra_dim; var weightMap, paramMappings, _a, extractMobilenetV1Params, extractPredictionLayerParams, extra_dim, params;
return __generator$1(this, function (_b) { return __generator$1(this, function (_b) {
switch (_b.label) { switch (_b.label) {
case 0: return [4 /*yield*/, loadWeightMap(uri, DEFAULT_MODEL_NAME)]; case 0: return [4 /*yield*/, loadWeightMap(uri, DEFAULT_MODEL_NAME)];
case 1: case 1:
weightMap = _b.sent(); weightMap = _b.sent();
_a = extractorsFactory$1(weightMap), extractMobilenetV1Params = _a.extractMobilenetV1Params, extractPredictionLayerParams = _a.extractPredictionLayerParams; paramMappings = [];
_a = extractorsFactory$1(weightMap, paramMappings), extractMobilenetV1Params = _a.extractMobilenetV1Params, extractPredictionLayerParams = _a.extractPredictionLayerParams;
extra_dim = weightMap['Output/extra_dim']; extra_dim = weightMap['Output/extra_dim'];
paramMappings.push({ originalPath: 'Output/extra_dim', paramPath: 'output_layer/extra_dim' });
if (!isTensor3D(extra_dim)) { if (!isTensor3D(extra_dim)) {
throw new Error("expected weightMap['Output/extra_dim'] to be a Tensor3D, instead have " + extra_dim); throw new Error("expected weightMap['Output/extra_dim'] to be a Tensor3D, instead have " + extra_dim);
} }
return [2 /*return*/, { params = {
mobilenetv1_params: extractMobilenetV1Params(), mobilenetv1: extractMobilenetV1Params(),
prediction_layer_params: extractPredictionLayerParams(), prediction_layer: extractPredictionLayerParams(),
output_layer_params: { output_layer: {
extra_dim: extra_dim extra_dim: extra_dim
} }
}]; };
disposeUnusedWeightTensors(weightMap, paramMappings);
return [2 /*return*/, { params: params, paramMappings: paramMappings }];
} }
}); });
}); });
...@@ -1295,12 +1431,27 @@ ...@@ -1295,12 +1431,27 @@
function mobileNetV1(x, params) { function mobileNetV1(x, params) {
return tidy(function () { return tidy(function () {
var conv11 = null; var conv11 = null;
var out = pointwiseConvLayer(x, params.conv_0_params, [2, 2]); var out = pointwiseConvLayer(x, params.conv_0, [2, 2]);
params.conv_pair_params.forEach(function (param, i) { var convPairParams = [
params.conv_1,
params.conv_2,
params.conv_3,
params.conv_4,
params.conv_5,
params.conv_6,
params.conv_7,
params.conv_8,
params.conv_9,
params.conv_10,
params.conv_11,
params.conv_12,
params.conv_13
];
convPairParams.forEach(function (param, i) {
var layerIdx = i + 1; var layerIdx = i + 1;
var depthwiseConvStrides = getStridesForLayerIdx(layerIdx); var depthwiseConvStrides = getStridesForLayerIdx(layerIdx);
out = depthwiseConvLayer(out, param.depthwise_conv_params, depthwiseConvStrides); out = depthwiseConvLayer(out, param.depthwise_conv, depthwiseConvStrides);
out = pointwiseConvLayer(out, param.pointwise_conv_params, [1, 1]); out = pointwiseConvLayer(out, param.pointwise_conv, [1, 1]);
if (layerIdx === 11) { if (layerIdx === 11) {
conv11 = out; conv11 = out;
} }
...@@ -1426,8 +1577,8 @@ ...@@ -1426,8 +1577,8 @@
function boxPredictionLayer(x, params) { function boxPredictionLayer(x, params) {
return tidy(function () { return tidy(function () {
var batchSize = x.shape[0]; var batchSize = x.shape[0];
var boxPredictionEncoding = reshape(convLayer(x, params.box_encoding_predictor_params), [batchSize, -1, 1, 4]); var boxPredictionEncoding = reshape(convLayer(x, params.box_encoding_predictor), [batchSize, -1, 1, 4]);
var classPrediction = reshape(convLayer(x, params.class_predictor_params), [batchSize, -1, 3]); var classPrediction = reshape(convLayer(x, params.class_predictor), [batchSize, -1, 3]);
return { return {
boxPredictionEncoding: boxPredictionEncoding, boxPredictionEncoding: boxPredictionEncoding,
classPrediction: classPrediction classPrediction: classPrediction
...@@ -1437,20 +1588,20 @@ ...@@ -1437,20 +1588,20 @@
function predictionLayer(x, conv11, params) { function predictionLayer(x, conv11, params) {
return tidy(function () { return tidy(function () {
var conv0 = pointwiseConvLayer(x, params.conv_0_params, [1, 1]); var conv0 = pointwiseConvLayer(x, params.conv_0, [1, 1]);
var conv1 = pointwiseConvLayer(conv0, params.conv_1_params, [2, 2]); var conv1 = pointwiseConvLayer(conv0, params.conv_1, [2, 2]);
var conv2 = pointwiseConvLayer(conv1, params.conv_2_params, [1, 1]); var conv2 = pointwiseConvLayer(conv1, params.conv_2, [1, 1]);
var conv3 = pointwiseConvLayer(conv2, params.conv_3_params, [2, 2]); var conv3 = pointwiseConvLayer(conv2, params.conv_3, [2, 2]);
var conv4 = pointwiseConvLayer(conv3, params.conv_4_params, [1, 1]); var conv4 = pointwiseConvLayer(conv3, params.conv_4, [1, 1]);
var conv5 = pointwiseConvLayer(conv4, params.conv_5_params, [2, 2]); var conv5 = pointwiseConvLayer(conv4, params.conv_5, [2, 2]);
var conv6 = pointwiseConvLayer(conv5, params.conv_6_params, [1, 1]); var conv6 = pointwiseConvLayer(conv5, params.conv_6, [1, 1]);
var conv7 = pointwiseConvLayer(conv6, params.conv_7_params, [2, 2]); var conv7 = pointwiseConvLayer(conv6, params.conv_7, [2, 2]);
var boxPrediction0 = boxPredictionLayer(conv11, params.box_predictor_0_params); var boxPrediction0 = boxPredictionLayer(conv11, params.box_predictor_0);
var boxPrediction1 = boxPredictionLayer(x, params.box_predictor_1_params); var boxPrediction1 = boxPredictionLayer(x, params.box_predictor_1);
var boxPrediction2 = boxPredictionLayer(conv1, params.box_predictor_2_params); var boxPrediction2 = boxPredictionLayer(conv1, params.box_predictor_2);
var boxPrediction3 = boxPredictionLayer(conv3, params.box_predictor_3_params); var boxPrediction3 = boxPredictionLayer(conv3, params.box_predictor_3);
var boxPrediction4 = boxPredictionLayer(conv5, params.box_predictor_4_params); var boxPrediction4 = boxPredictionLayer(conv5, params.box_predictor_4);
var boxPrediction5 = boxPredictionLayer(conv7, params.box_predictor_5_params); var boxPrediction5 = boxPredictionLayer(conv7, params.box_predictor_5);
var boxPredictions = concat([ var boxPredictions = concat([
boxPrediction0.boxPredictionEncoding, boxPrediction0.boxPredictionEncoding,
boxPrediction1.boxPredictionEncoding, boxPrediction1.boxPredictionEncoding,
...@@ -1474,45 +1625,22 @@ ...@@ -1474,45 +1625,22 @@
}); });
} }
var FaceDetectionNet = /** @class */ (function () { var FaceDetectionNet = /** @class */ (function (_super) {
__extends$1(FaceDetectionNet, _super);
function FaceDetectionNet() { function FaceDetectionNet() {
return _super.call(this, 'FaceDetectionNet') || this;
} }
FaceDetectionNet.prototype.load = function (weightsOrUrl) {
return __awaiter$1(this, void 0, void 0, function () {
var _a;
return __generator$1(this, function (_b) {
switch (_b.label) {
case 0:
if (weightsOrUrl instanceof Float32Array) {
this.extractWeights(weightsOrUrl);
return [2 /*return*/];
}
if (weightsOrUrl && typeof weightsOrUrl !== 'string') {
throw new Error('FaceDetectionNet.load - expected model uri, or weights as Float32Array');
}
_a = this;
return [4 /*yield*/, loadQuantizedParams(weightsOrUrl)];
case 1:
_a._params = _b.sent();
return [2 /*return*/];
}
});
});
};
FaceDetectionNet.prototype.extractWeights = function (weights) {
this._params = extractParams(weights);
};
FaceDetectionNet.prototype.forwardInput = function (input) { FaceDetectionNet.prototype.forwardInput = function (input) {
var _this = this; var params = this.params;
if (!this._params) { if (!params) {
throw new Error('FaceDetectionNet - load model before inference'); throw new Error('FaceDetectionNet - load model before inference');
} }
return tidy(function () { return tidy(function () {
var batchTensor = input.toBatchTensor(512, false); var batchTensor = input.toBatchTensor(512, false);
var x = sub(mul(batchTensor, scalar(0.007843137718737125)), scalar(1)); var x = sub(mul(batchTensor, scalar(0.007843137718737125)), scalar(1));
var features = mobileNetV1(x, _this._params.mobilenetv1_params); var features = mobileNetV1(x, params.mobilenetv1);
var _a = predictionLayer(features.out, features.conv11, _this._params.prediction_layer_params), boxPredictions = _a.boxPredictions, classPredictions = _a.classPredictions; var _a = predictionLayer(features.out, features.conv11, params.prediction_layer), boxPredictions = _a.boxPredictions, classPredictions = _a.classPredictions;
return outputLayer(boxPredictions, classPredictions, _this._params.output_layer_params); return outputLayer(boxPredictions, classPredictions, params.output_layer);
}); });
}; };
FaceDetectionNet.prototype.forward = function (input) { FaceDetectionNet.prototype.forward = function (input) {
...@@ -1575,8 +1703,14 @@ ...@@ -1575,8 +1703,14 @@
}); });
}); });
}; };
FaceDetectionNet.prototype.loadQuantizedParams = function (uri) {
return loadQuantizedParams(uri);
};
FaceDetectionNet.prototype.extractParams = function (weights) {
return extractParams(weights);
};
return FaceDetectionNet; return FaceDetectionNet;
}()); }(NeuralNetwork));
function faceDetectionNet(weights) { function faceDetectionNet(weights) {
var net = new FaceDetectionNet(); var net = new FaceDetectionNet();
...@@ -1584,103 +1718,15 @@ ...@@ -1584,103 +1718,15 @@
return net; return net;
} }
var NeuralNetwork = /** @class */ (function () { function extractParams$1(weights) {
function NeuralNetwork() { var paramMappings = [];
this._params = undefined; var _a = extractWeightsFactory(weights), extractWeights = _a.extractWeights, getRemainingWeights = _a.getRemainingWeights;
this._paramMappings = []; function extractConvParams(channelsIn, channelsOut, filterSize, mappedPrefix) {
}
Object.defineProperty(NeuralNetwork.prototype, "params", {
get: function () {
return this._params;
},
enumerable: true,
configurable: true
});
Object.defineProperty(NeuralNetwork.prototype, "paramMappings", {
get: function () {
return this._paramMappings;
},
enumerable: true,
configurable: true
});
NeuralNetwork.prototype.getParamFromPath = function (paramPath) {
var _a = this.traversePropertyPath(paramPath), obj = _a.obj, objProp = _a.objProp;
return obj[objProp];
};
NeuralNetwork.prototype.reassignParamFromPath = function (paramPath, tensor$$1) {
var _a = this.traversePropertyPath(paramPath), obj = _a.obj, objProp = _a.objProp;
obj[objProp].dispose();
obj[objProp] = tensor$$1;
};
NeuralNetwork.prototype.getParamList = function () {
var _this = this;
return this._paramMappings.map(function (_a) {
var paramPath = _a.paramPath;
return ({
path: paramPath,
tensor: _this.getParamFromPath(paramPath)
});
});
};
NeuralNetwork.prototype.getTrainableParams = function () {
return this.getParamList().filter(function (param) { return param.tensor instanceof Variable; });
};
NeuralNetwork.prototype.getFrozenParams = function () {
return this.getParamList().filter(function (param) { return !(param.tensor instanceof Variable); });
};
NeuralNetwork.prototype.variable = function () {
var _this = this;
this.getFrozenParams().forEach(function (_a) {
var path = _a.path, tensor$$1 = _a.tensor;
_this.reassignParamFromPath(path, variable(tensor$$1));
});
};
NeuralNetwork.prototype.freeze = function () {
var _this = this;
this.getTrainableParams().forEach(function (_a) {
var path = _a.path, tensor$$1 = _a.tensor;
_this.reassignParamFromPath(path, tensor(tensor$$1));
});
};
NeuralNetwork.prototype.dispose = function () {
this.getParamList().forEach(function (param) { return param.tensor.dispose(); });
this._params = undefined;
};
NeuralNetwork.prototype.traversePropertyPath = function (paramPath) {
if (!this.params) {
throw new Error("traversePropertyPath - model has no loaded params");
}
var result = paramPath.split('/').reduce(function (res, objProp) {
if (!res.nextObj.hasOwnProperty(objProp)) {
throw new Error("traversePropertyPath - object does not have property " + objProp + ", for path " + paramPath);
}
return { obj: res.nextObj, objProp: objProp, nextObj: res.nextObj[objProp] };
}, { nextObj: this.params });
var obj = result.obj, objProp = result.objProp;
if (!obj || !objProp || !(obj[objProp] instanceof Tensor)) {
throw new Error("traversePropertyPath - parameter is not a tensor, for path " + paramPath);
}
return { obj: obj, objProp: objProp };
};
return NeuralNetwork;
}());
function extractConvParamsFactory(extractWeights, paramMappings) {
return function (channelsIn, channelsOut, filterSize, mappedPrefix) {
var filters = tensor4d(extractWeights(channelsIn * channelsOut * filterSize * filterSize), [filterSize, filterSize, channelsIn, channelsOut]); var filters = tensor4d(extractWeights(channelsIn * channelsOut * filterSize * filterSize), [filterSize, filterSize, channelsIn, channelsOut]);
var bias = tensor1d(extractWeights(channelsOut)); var bias = tensor1d(extractWeights(channelsOut));
paramMappings.push({ paramPath: mappedPrefix + "/filters" }, { paramPath: mappedPrefix + "/bias" }); paramMappings.push({ paramPath: mappedPrefix + "/filters" }, { paramPath: mappedPrefix + "/bias" });
return { return { filters: filters, bias: bias };
filters: filters,
bias: bias
};
};
} }
function extractParams$1(weights) {
var paramMappings = [];
var _a = extractWeightsFactory(weights), extractWeights = _a.extractWeights, getRemainingWeights = _a.getRemainingWeights;
var extractConvParams = extractConvParamsFactory(extractWeights, paramMappings);
function extractFcParams(channelsIn, channelsOut, mappedPrefix) { function extractFcParams(channelsIn, channelsOut, mappedPrefix) {
var fc_weights = tensor2d(extractWeights(channelsIn * channelsOut), [channelsIn, channelsOut]); var fc_weights = tensor2d(extractWeights(channelsIn * channelsOut), [channelsIn, channelsOut]);
var fc_bias = tensor1d(extractWeights(channelsOut)); var fc_bias = tensor1d(extractWeights(channelsOut));
...@@ -1690,32 +1736,32 @@ ...@@ -1690,32 +1736,32 @@
bias: fc_bias bias: fc_bias
}; };
} }
var conv0_params = extractConvParams(3, 32, 3, 'conv0_params'); var conv0 = extractConvParams(3, 32, 3, 'conv0');
var conv1_params = extractConvParams(32, 64, 3, 'conv1_params'); var conv1 = extractConvParams(32, 64, 3, 'conv1');
var conv2_params = extractConvParams(64, 64, 3, 'conv2_params'); var conv2 = extractConvParams(64, 64, 3, 'conv2');
var conv3_params = extractConvParams(64, 64, 3, 'conv3_params'); var conv3 = extractConvParams(64, 64, 3, 'conv3');
var conv4_params = extractConvParams(64, 64, 3, 'conv4_params'); var conv4 = extractConvParams(64, 64, 3, 'conv4');
var conv5_params = extractConvParams(64, 128, 3, 'conv5_params'); var conv5 = extractConvParams(64, 128, 3, 'conv5');
var conv6_params = extractConvParams(128, 128, 3, 'conv6_params'); var conv6 = extractConvParams(128, 128, 3, 'conv6');
var conv7_params = extractConvParams(128, 256, 3, 'conv7_params'); var conv7 = extractConvParams(128, 256, 3, 'conv7');
var fc0_params = extractFcParams(6400, 1024, 'fc0_params'); var fc0 = extractFcParams(6400, 1024, 'fc0');
var fc1_params = extractFcParams(1024, 136, 'fc1_params'); var fc1 = extractFcParams(1024, 136, 'fc1');
if (getRemainingWeights().length !== 0) { if (getRemainingWeights().length !== 0) {
throw new Error("weights remaing after extract: " + getRemainingWeights().length); throw new Error("weights remaing after extract: " + getRemainingWeights().length);
} }
return { return {
paramMappings: paramMappings, paramMappings: paramMappings,
params: { params: {
conv0_params: conv0_params, conv0: conv0,
conv1_params: conv1_params, conv1: conv1,
conv2_params: conv2_params, conv2: conv2,
conv3_params: conv3_params, conv3: conv3,
conv4_params: conv4_params, conv4: conv4,
conv5_params: conv5_params, conv5: conv5,
conv6_params: conv6_params, conv6: conv6,
conv7_params: conv7_params, conv7: conv7,
fc0_params: fc0_params, fc0: fc0,
fc1_params: fc1_params fc1: fc1
} }
}; };
} }
...@@ -1826,33 +1872,18 @@ ...@@ -1826,33 +1872,18 @@
}); });
} }
function extractWeightEntry(weightMap, path, paramRank) {
var tensor = weightMap[path];
if (!isTensor(tensor, paramRank)) {
throw new Error("expected weightMap[" + path + "] to be a Tensor" + paramRank + "D, instead have " + tensor);
}
return { path: path, tensor: tensor };
}
var DEFAULT_MODEL_NAME$1 = 'face_landmark_68_model'; var DEFAULT_MODEL_NAME$1 = 'face_landmark_68_model';
function extractorsFactory$2(weightMap, paramMappings) { function extractorsFactory$2(weightMap, paramMappings) {
var extractWeightEntry = extractWeightEntryFactory(weightMap, paramMappings);
function extractConvParams(prefix, mappedPrefix) { function extractConvParams(prefix, mappedPrefix) {
var filtersEntry = extractWeightEntry(weightMap, prefix + "/kernel", 4); var filters = extractWeightEntry(prefix + "/kernel", 4, mappedPrefix + "/filters");
var biasEntry = extractWeightEntry(weightMap, prefix + "/bias", 1); var bias = extractWeightEntry(prefix + "/bias", 1, mappedPrefix + "/bias");
paramMappings.push({ originalPath: filtersEntry.path, paramPath: mappedPrefix + "/filters" }, { originalPath: biasEntry.path, paramPath: mappedPrefix + "/bias" }); return { filters: filters, bias: bias };
return {
filters: filtersEntry.tensor,
bias: biasEntry.tensor
};
} }
function extractFcParams(prefix, mappedPrefix) { function extractFcParams(prefix, mappedPrefix) {
var weightsEntry = extractWeightEntry(weightMap, prefix + "/kernel", 2); var weights = extractWeightEntry(prefix + "/kernel", 2, mappedPrefix + "/weights");
var biasEntry = extractWeightEntry(weightMap, prefix + "/bias", 1); var bias = extractWeightEntry(prefix + "/bias", 1, mappedPrefix + "/bias");
paramMappings.push({ originalPath: weightsEntry.path, paramPath: mappedPrefix + "/weights" }, { originalPath: biasEntry.path, paramPath: mappedPrefix + "/bias" }); return { weights: weights, bias: bias };
return {
weights: weightsEntry.tensor,
bias: biasEntry.tensor
};
} }
return { return {
extractConvParams: extractConvParams, extractConvParams: extractConvParams,
...@@ -1870,17 +1901,18 @@ ...@@ -1870,17 +1901,18 @@
paramMappings = []; paramMappings = [];
_a = extractorsFactory$2(weightMap, paramMappings), extractConvParams = _a.extractConvParams, extractFcParams = _a.extractFcParams; _a = extractorsFactory$2(weightMap, paramMappings), extractConvParams = _a.extractConvParams, extractFcParams = _a.extractFcParams;
params = { params = {
conv0_params: extractConvParams('conv2d_0', 'conv0_params'), conv0: extractConvParams('conv2d_0', 'conv0'),
conv1_params: extractConvParams('conv2d_1', 'conv1_params'), conv1: extractConvParams('conv2d_1', 'conv1'),
conv2_params: extractConvParams('conv2d_2', 'conv2_params'), conv2: extractConvParams('conv2d_2', 'conv2'),
conv3_params: extractConvParams('conv2d_3', 'conv3_params'), conv3: extractConvParams('conv2d_3', 'conv3'),
conv4_params: extractConvParams('conv2d_4', 'conv4_params'), conv4: extractConvParams('conv2d_4', 'conv4'),
conv5_params: extractConvParams('conv2d_5', 'conv5_params'), conv5: extractConvParams('conv2d_5', 'conv5'),
conv6_params: extractConvParams('conv2d_6', 'conv6_params'), conv6: extractConvParams('conv2d_6', 'conv6'),
conv7_params: extractConvParams('conv2d_7', 'conv7_params'), conv7: extractConvParams('conv2d_7', 'conv7'),
fc0_params: extractFcParams('dense', 'fc0_params'), fc0: extractFcParams('dense', 'fc0'),
fc1_params: extractFcParams('logits', 'fc1_params') fc1: extractFcParams('logits', 'fc1')
}; };
disposeUnusedWeightTensors(weightMap, paramMappings);
return [2 /*return*/, { params: params, paramMappings: paramMappings }]; return [2 /*return*/, { params: params, paramMappings: paramMappings }];
} }
}); });
...@@ -1897,57 +1929,29 @@ ...@@ -1897,57 +1929,29 @@
var FaceLandmarkNet = /** @class */ (function (_super) { var FaceLandmarkNet = /** @class */ (function (_super) {
__extends$1(FaceLandmarkNet, _super); __extends$1(FaceLandmarkNet, _super);
function FaceLandmarkNet() { function FaceLandmarkNet() {
return _super !== null && _super.apply(this, arguments) || this; return _super.call(this, 'FaceLandmarkNet') || this;
}
FaceLandmarkNet.prototype.load = function (weightsOrUrl) {
return __awaiter$1(this, void 0, void 0, function () {
var _a, paramMappings, params;
return __generator$1(this, function (_b) {
switch (_b.label) {
case 0:
if (weightsOrUrl instanceof Float32Array) {
this.extractWeights(weightsOrUrl);
return [2 /*return*/];
}
if (weightsOrUrl && typeof weightsOrUrl !== 'string') {
throw new Error('FaceLandmarkNet.load - expected model uri, or weights as Float32Array');
} }
return [4 /*yield*/, loadQuantizedParams$1(weightsOrUrl)];
case 1:
_a = _b.sent(), paramMappings = _a.paramMappings, params = _a.params;
this._paramMappings = paramMappings;
this._params = params;
return [2 /*return*/];
}
});
});
};
FaceLandmarkNet.prototype.extractWeights = function (weights) {
var _a = extractParams$1(weights), paramMappings = _a.paramMappings, params = _a.params;
this._paramMappings = paramMappings;
this._params = params;
};
FaceLandmarkNet.prototype.forwardInput = function (input) { FaceLandmarkNet.prototype.forwardInput = function (input) {
var params = this._params; var params = this.params;
if (!params) { if (!params) {
throw new Error('FaceLandmarkNet - load model before inference'); throw new Error('FaceLandmarkNet - load model before inference');
} }
return tidy(function () { return tidy(function () {
var batchTensor = input.toBatchTensor(128, true); var batchTensor = input.toBatchTensor(128, true);
var out = conv(batchTensor, params.conv0_params); var out = conv(batchTensor, params.conv0);
out = maxPool$1(out); out = maxPool$1(out);
out = conv(out, params.conv1_params); out = conv(out, params.conv1);
out = conv(out, params.conv2_params); out = conv(out, params.conv2);
out = maxPool$1(out); out = maxPool$1(out);
out = conv(out, params.conv3_params); out = conv(out, params.conv3);
out = conv(out, params.conv4_params); out = conv(out, params.conv4);
out = maxPool$1(out); out = maxPool$1(out);
out = conv(out, params.conv5_params); out = conv(out, params.conv5);
out = conv(out, params.conv6_params); out = conv(out, params.conv6);
out = maxPool$1(out, [1, 1]); out = maxPool$1(out, [1, 1]);
out = conv(out, params.conv7_params); out = conv(out, params.conv7);
var fc0 = relu(fullyConnectedLayer(out.as2D(out.shape[0], -1), params.fc0_params)); var fc0 = relu(fullyConnectedLayer(out.as2D(out.shape[0], -1), params.fc0));
var fc1 = fullyConnectedLayer(fc0, params.fc1_params); var fc1 = fullyConnectedLayer(fc0, params.fc1);
var createInterleavedTensor = function (fillX, fillY) { var createInterleavedTensor = function (fillX, fillY) {
return stack([ return stack([
fill([68], fillX), fill([68], fillX),
...@@ -2022,6 +2026,12 @@ ...@@ -2022,6 +2026,12 @@
}); });
}); });
}; };
FaceLandmarkNet.prototype.loadQuantizedParams = function (uri) {
return loadQuantizedParams$1(uri);
};
FaceLandmarkNet.prototype.extractParams = function (weights) {
return extractParams$1(weights);
};
return FaceLandmarkNet; return FaceLandmarkNet;
}(NeuralNetwork)); }(NeuralNetwork));
...@@ -2053,43 +2063,40 @@ ...@@ -2053,43 +2063,40 @@
return convLayer$1(x, params, [2, 2], true, 'valid'); return convLayer$1(x, params, [2, 2], true, 'valid');
} }
function extractorsFactory$3(extractWeights) { function extractorsFactory$3(extractWeights, paramMappings) {
function extractFilterValues(numFilterValues, numFilters, filterSize) { function extractFilterValues(numFilterValues, numFilters, filterSize) {
var weights = extractWeights(numFilterValues); var weights = extractWeights(numFilterValues);
var depth = weights.length / (numFilters * filterSize * filterSize); var depth = weights.length / (numFilters * filterSize * filterSize);
if (isFloat(depth)) { if (isFloat(depth)) {
throw new Error("depth has to be an integer: " + depth + ", weights.length: " + weights.length + ", numFilters: " + numFilters + ", filterSize: " + filterSize); throw new Error("depth has to be an integer: " + depth + ", weights.length: " + weights.length + ", numFilters: " + numFilters + ", filterSize: " + filterSize);
} }
return transpose(tensor4d(weights, [numFilters, depth, filterSize, filterSize]), [2, 3, 1, 0]); return tidy(function () { return transpose(tensor4d(weights, [numFilters, depth, filterSize, filterSize]), [2, 3, 1, 0]); });
} }
function extractScaleLayerParams(numWeights) { function extractConvParams(numFilterValues, numFilters, filterSize, mappedPrefix) {
var filters = extractFilterValues(numFilterValues, numFilters, filterSize);
var bias = tensor1d(extractWeights(numFilters));
paramMappings.push({ paramPath: mappedPrefix + "/filters" }, { paramPath: mappedPrefix + "/bias" });
return { filters: filters, bias: bias };
}
function extractScaleLayerParams(numWeights, mappedPrefix) {
var weights = tensor1d(extractWeights(numWeights)); var weights = tensor1d(extractWeights(numWeights));
var biases = tensor1d(extractWeights(numWeights)); var biases = tensor1d(extractWeights(numWeights));
paramMappings.push({ paramPath: mappedPrefix + "/weights" }, { paramPath: mappedPrefix + "/biases" });
return { return {
weights: weights, weights: weights,
biases: biases biases: biases
}; };
} }
function extractConvLayerParams(numFilterValues, numFilters, filterSize) { function extractConvLayerParams(numFilterValues, numFilters, filterSize, mappedPrefix) {
var conv_filters = extractFilterValues(numFilterValues, numFilters, filterSize); var conv = extractConvParams(numFilterValues, numFilters, filterSize, mappedPrefix + "/conv");
var conv_bias = tensor1d(extractWeights(numFilters)); var scale = extractScaleLayerParams(numFilters, mappedPrefix + "/scale");
var scale = extractScaleLayerParams(numFilters); return { conv: conv, scale: scale };
return {
conv: {
filters: conv_filters,
bias: conv_bias
},
scale: scale
};
} }
function extractResidualLayerParams(numFilterValues, numFilters, filterSize, isDown) { function extractResidualLayerParams(numFilterValues, numFilters, filterSize, mappedPrefix, isDown) {
if (isDown === void 0) { isDown = false; } if (isDown === void 0) { isDown = false; }
var conv1 = extractConvLayerParams((isDown ? 0.5 : 1) * numFilterValues, numFilters, filterSize); var conv1 = extractConvLayerParams((isDown ? 0.5 : 1) * numFilterValues, numFilters, filterSize, mappedPrefix + "/conv1");
var conv2 = extractConvLayerParams(numFilterValues, numFilters, filterSize); var conv2 = extractConvLayerParams(numFilterValues, numFilters, filterSize, mappedPrefix + "/conv2");
return { return { conv1: conv1, conv2: conv2 };
conv1: conv1,
conv2: conv2
};
} }
return { return {
extractConvLayerParams: extractConvLayerParams, extractConvLayerParams: extractConvLayerParams,
...@@ -2098,27 +2105,29 @@ ...@@ -2098,27 +2105,29 @@
} }
function extractParams$2(weights) { function extractParams$2(weights) {
var _a = extractWeightsFactory(weights), extractWeights = _a.extractWeights, getRemainingWeights = _a.getRemainingWeights; var _a = extractWeightsFactory(weights), extractWeights = _a.extractWeights, getRemainingWeights = _a.getRemainingWeights;
var _b = extractorsFactory$3(extractWeights), extractConvLayerParams = _b.extractConvLayerParams, extractResidualLayerParams = _b.extractResidualLayerParams; var paramMappings = [];
var conv32_down = extractConvLayerParams(4704, 32, 7); var _b = extractorsFactory$3(extractWeights, paramMappings), extractConvLayerParams = _b.extractConvLayerParams, extractResidualLayerParams = _b.extractResidualLayerParams;
var conv32_1 = extractResidualLayerParams(9216, 32, 3); var conv32_down = extractConvLayerParams(4704, 32, 7, 'conv32_down');
var conv32_2 = extractResidualLayerParams(9216, 32, 3); var conv32_1 = extractResidualLayerParams(9216, 32, 3, 'conv32_1');
var conv32_3 = extractResidualLayerParams(9216, 32, 3); var conv32_2 = extractResidualLayerParams(9216, 32, 3, 'conv32_2');
var conv64_down = extractResidualLayerParams(36864, 64, 3, true); var conv32_3 = extractResidualLayerParams(9216, 32, 3, 'conv32_3');
var conv64_1 = extractResidualLayerParams(36864, 64, 3); var conv64_down = extractResidualLayerParams(36864, 64, 3, 'conv64_down', true);
var conv64_2 = extractResidualLayerParams(36864, 64, 3); var conv64_1 = extractResidualLayerParams(36864, 64, 3, 'conv64_1');
var conv64_3 = extractResidualLayerParams(36864, 64, 3); var conv64_2 = extractResidualLayerParams(36864, 64, 3, 'conv64_2');
var conv128_down = extractResidualLayerParams(147456, 128, 3, true); var conv64_3 = extractResidualLayerParams(36864, 64, 3, 'conv64_3');
var conv128_1 = extractResidualLayerParams(147456, 128, 3); var conv128_down = extractResidualLayerParams(147456, 128, 3, 'conv128_down', true);
var conv128_2 = extractResidualLayerParams(147456, 128, 3); var conv128_1 = extractResidualLayerParams(147456, 128, 3, 'conv128_1');
var conv256_down = extractResidualLayerParams(589824, 256, 3, true); var conv128_2 = extractResidualLayerParams(147456, 128, 3, 'conv128_2');
var conv256_1 = extractResidualLayerParams(589824, 256, 3); var conv256_down = extractResidualLayerParams(589824, 256, 3, 'conv256_down', true);
var conv256_2 = extractResidualLayerParams(589824, 256, 3); var conv256_1 = extractResidualLayerParams(589824, 256, 3, 'conv256_1');
var conv256_down_out = extractResidualLayerParams(589824, 256, 3); var conv256_2 = extractResidualLayerParams(589824, 256, 3, 'conv256_2');
var fc = transpose(tensor2d(extractWeights(256 * 128), [128, 256]), [1, 0]); var conv256_down_out = extractResidualLayerParams(589824, 256, 3, 'conv256_down_out');
var fc = tidy(function () { return transpose(tensor2d(extractWeights(256 * 128), [128, 256]), [1, 0]); });
paramMappings.push({ paramPath: "fc" });
if (getRemainingWeights().length !== 0) { if (getRemainingWeights().length !== 0) {
throw new Error("weights remaing after extract: " + getRemainingWeights().length); throw new Error("weights remaing after extract: " + getRemainingWeights().length);
} }
return { var params = {
conv32_down: conv32_down, conv32_down: conv32_down,
conv32_1: conv32_1, conv32_1: conv32_1,
conv32_2: conv32_2, conv32_2: conv32_2,
...@@ -2136,38 +2145,22 @@ ...@@ -2136,38 +2145,22 @@
conv256_down_out: conv256_down_out, conv256_down_out: conv256_down_out,
fc: fc fc: fc
}; };
return { params: params, paramMappings: paramMappings };
} }
var DEFAULT_MODEL_NAME$2 = 'face_recognition_model'; var DEFAULT_MODEL_NAME$2 = 'face_recognition_model';
function extractorsFactory$4(weightMap) { function extractorsFactory$4(weightMap, paramMappings) {
var extractWeightEntry = extractWeightEntryFactory(weightMap, paramMappings);
function extractScaleLayerParams(prefix) { function extractScaleLayerParams(prefix) {
var params = { var weights = extractWeightEntry(prefix + "/scale/weights", 1);
weights: weightMap[prefix + "/scale/weights"], var biases = extractWeightEntry(prefix + "/scale/biases", 1);
biases: weightMap[prefix + "/scale/biases"] return { weights: weights, biases: biases };
};
if (!isTensor1D(params.weights)) {
throw new Error("expected weightMap[" + prefix + "/scale/weights] to be a Tensor1D, instead have " + params.weights);
}
if (!isTensor1D(params.biases)) {
throw new Error("expected weightMap[" + prefix + "/scale/biases] to be a Tensor1D, instead have " + params.biases);
}
return params;
} }
function extractConvLayerParams(prefix) { function extractConvLayerParams(prefix) {
var params = { var filters = extractWeightEntry(prefix + "/conv/filters", 4);
filters: weightMap[prefix + "/conv/filters"], var bias = extractWeightEntry(prefix + "/conv/bias", 1);
bias: weightMap[prefix + "/conv/bias"] var scale = extractScaleLayerParams(prefix);
}; return { conv: { filters: filters, bias: bias }, scale: scale };
if (!isTensor4D(params.filters)) {
throw new Error("expected weightMap[" + prefix + "/conv/filters] to be a Tensor1D, instead have " + params.filters);
}
if (!isTensor1D(params.bias)) {
throw new Error("expected weightMap[" + prefix + "/conv/bias] to be a Tensor1D, instead have " + params.bias);
}
return {
conv: params,
scale: extractScaleLayerParams(prefix)
};
} }
function extractResidualLayerParams(prefix) { function extractResidualLayerParams(prefix) {
return { return {
...@@ -2182,13 +2175,14 @@ ...@@ -2182,13 +2175,14 @@
} }
function loadQuantizedParams$2(uri) { function loadQuantizedParams$2(uri) {
return __awaiter$1(this, void 0, void 0, function () { return __awaiter$1(this, void 0, void 0, function () {
var weightMap, _a, extractConvLayerParams, extractResidualLayerParams, conv32_down, conv32_1, conv32_2, conv32_3, conv64_down, conv64_1, conv64_2, conv64_3, conv128_down, conv128_1, conv128_2, conv256_down, conv256_1, conv256_2, conv256_down_out, fc; var weightMap, paramMappings, _a, extractConvLayerParams, extractResidualLayerParams, conv32_down, conv32_1, conv32_2, conv32_3, conv64_down, conv64_1, conv64_2, conv64_3, conv128_down, conv128_1, conv128_2, conv256_down, conv256_1, conv256_2, conv256_down_out, fc, params;
return __generator$1(this, function (_b) { return __generator$1(this, function (_b) {
switch (_b.label) { switch (_b.label) {
case 0: return [4 /*yield*/, loadWeightMap(uri, DEFAULT_MODEL_NAME$2)]; case 0: return [4 /*yield*/, loadWeightMap(uri, DEFAULT_MODEL_NAME$2)];
case 1: case 1:
weightMap = _b.sent(); weightMap = _b.sent();
_a = extractorsFactory$4(weightMap), extractConvLayerParams = _a.extractConvLayerParams, extractResidualLayerParams = _a.extractResidualLayerParams; paramMappings = [];
_a = extractorsFactory$4(weightMap, paramMappings), extractConvLayerParams = _a.extractConvLayerParams, extractResidualLayerParams = _a.extractResidualLayerParams;
conv32_down = extractConvLayerParams('conv32_down'); conv32_down = extractConvLayerParams('conv32_down');
conv32_1 = extractResidualLayerParams('conv32_1'); conv32_1 = extractResidualLayerParams('conv32_1');
conv32_2 = extractResidualLayerParams('conv32_2'); conv32_2 = extractResidualLayerParams('conv32_2');
...@@ -2205,10 +2199,11 @@ ...@@ -2205,10 +2199,11 @@
conv256_2 = extractResidualLayerParams('conv256_2'); conv256_2 = extractResidualLayerParams('conv256_2');
conv256_down_out = extractResidualLayerParams('conv256_down_out'); conv256_down_out = extractResidualLayerParams('conv256_down_out');
fc = weightMap['fc']; fc = weightMap['fc'];
paramMappings.push({ originalPath: 'fc', paramPath: 'fc' });
if (!isTensor2D(fc)) { if (!isTensor2D(fc)) {
throw new Error("expected weightMap[fc] to be a Tensor2D, instead have " + fc); throw new Error("expected weightMap[fc] to be a Tensor2D, instead have " + fc);
} }
return [2 /*return*/, { params = {
conv32_down: conv32_down, conv32_down: conv32_down,
conv32_1: conv32_1, conv32_1: conv32_1,
conv32_2: conv32_2, conv32_2: conv32_2,
...@@ -2225,7 +2220,9 @@ ...@@ -2225,7 +2220,9 @@
conv256_2: conv256_2, conv256_2: conv256_2,
conv256_down_out: conv256_down_out, conv256_down_out: conv256_down_out,
fc: fc fc: fc
}]; };
disposeUnusedWeightTensors(weightMap, paramMappings);
return [2 /*return*/, { params: params, paramMappings: paramMappings }];
} }
}); });
}); });
...@@ -2271,60 +2268,37 @@ ...@@ -2271,60 +2268,37 @@
return out; return out;
} }
var FaceRecognitionNet = /** @class */ (function () { var FaceRecognitionNet = /** @class */ (function (_super) {
__extends$1(FaceRecognitionNet, _super);
function FaceRecognitionNet() { function FaceRecognitionNet() {
return _super.call(this, 'FaceRecognitionNet') || this;
} }
FaceRecognitionNet.prototype.load = function (weightsOrUrl) {
return __awaiter$1(this, void 0, void 0, function () {
var _a;
return __generator$1(this, function (_b) {
switch (_b.label) {
case 0:
if (weightsOrUrl instanceof Float32Array) {
this.extractWeights(weightsOrUrl);
return [2 /*return*/];
}
if (weightsOrUrl && typeof weightsOrUrl !== 'string') {
throw new Error('FaceLandmarkNet.load - expected model uri, or weights as Float32Array');
}
_a = this;
return [4 /*yield*/, loadQuantizedParams$2(weightsOrUrl)];
case 1:
_a._params = _b.sent();
return [2 /*return*/];
}
});
});
};
FaceRecognitionNet.prototype.extractWeights = function (weights) {
this._params = extractParams$2(weights);
};
FaceRecognitionNet.prototype.forwardInput = function (input) { FaceRecognitionNet.prototype.forwardInput = function (input) {
var _this = this; var params = this.params;
if (!this._params) { if (!params) {
throw new Error('FaceRecognitionNet - load model before inference'); throw new Error('FaceRecognitionNet - load model before inference');
} }
return tidy(function () { return tidy(function () {
var batchTensor = input.toBatchTensor(150, true); var batchTensor = input.toBatchTensor(150, true);
var normalized = normalize(batchTensor); var normalized = normalize(batchTensor);
var out = convDown(normalized, _this._params.conv32_down); var out = convDown(normalized, params.conv32_down);
out = maxPool(out, 3, 2, 'valid'); out = maxPool(out, 3, 2, 'valid');
out = residual(out, _this._params.conv32_1); out = residual(out, params.conv32_1);
out = residual(out, _this._params.conv32_2); out = residual(out, params.conv32_2);
out = residual(out, _this._params.conv32_3); out = residual(out, params.conv32_3);
out = residualDown(out, _this._params.conv64_down); out = residualDown(out, params.conv64_down);
out = residual(out, _this._params.conv64_1); out = residual(out, params.conv64_1);
out = residual(out, _this._params.conv64_2); out = residual(out, params.conv64_2);
out = residual(out, _this._params.conv64_3); out = residual(out, params.conv64_3);
out = residualDown(out, _this._params.conv128_down); out = residualDown(out, params.conv128_down);
out = residual(out, _this._params.conv128_1); out = residual(out, params.conv128_1);
out = residual(out, _this._params.conv128_2); out = residual(out, params.conv128_2);
out = residualDown(out, _this._params.conv256_down); out = residualDown(out, params.conv256_down);
out = residual(out, _this._params.conv256_1); out = residual(out, params.conv256_1);
out = residual(out, _this._params.conv256_2); out = residual(out, params.conv256_2);
out = residualDown(out, _this._params.conv256_down_out); out = residualDown(out, params.conv256_down_out);
var globalAvg = out.mean([1, 2]); var globalAvg = out.mean([1, 2]);
var fullyConnected = matMul(globalAvg, _this._params.fc); var fullyConnected = matMul(globalAvg, params.fc);
return fullyConnected; return fullyConnected;
}); });
}; };
...@@ -2362,8 +2336,14 @@ ...@@ -2362,8 +2336,14 @@
}); });
}); });
}; };
FaceRecognitionNet.prototype.loadQuantizedParams = function (uri) {
return loadQuantizedParams$2(uri);
};
FaceRecognitionNet.prototype.extractParams = function (weights) {
return extractParams$2(weights);
};
return FaceRecognitionNet; return FaceRecognitionNet;
}()); }(NeuralNetwork));
function faceRecognitionNet(weights) { function faceRecognitionNet(weights) {
var net = new FaceRecognitionNet(); var net = new FaceRecognitionNet();
......
dist/face-api.js.map
This source diff could not be displayed because it is too large. You can view the blob instead.
dist/face-api.min.js
This source diff could not be displayed because it is too large. You can view the blob instead.
src/Rect.ts
...@@ -32,6 +32,11 @@ export class Rect implements IRect { ...@@ -32,6 +32,11 @@ export class Rect implements IRect {
return new Rect(x, y, width, height) return new Rect(x, y, width, height)
} }
public pad(padX: number, padY: number): Rect {
let { x, y, width, height } = this
return new Rect(x - (padX / 2), y - (padY / 2), width + padX, height + padY)
}
public floor(): Rect { public floor(): Rect {
return new Rect( return new Rect(
Math.floor(this.x), Math.floor(this.x),
......
src/commons/NeuralNetwork.ts
...@@ -7,6 +7,8 @@ export class NeuralNetwork<TNetParams> { ...@@ -7,6 +7,8 @@ export class NeuralNetwork<TNetParams> {
protected _params: TNetParams | undefined = undefined protected _params: TNetParams | undefined = undefined
protected _paramMappings: ParamMapping[] = [] protected _paramMappings: ParamMapping[] = []
constructor(private _name: string) {}
public get params(): TNetParams | undefined { public get params(): TNetParams | undefined {
return this._params return this._params
} }
...@@ -53,11 +55,44 @@ export class NeuralNetwork<TNetParams> { ...@@ -53,11 +55,44 @@ export class NeuralNetwork<TNetParams> {
}) })
} }
public dispose() { public dispose(throwOnRedispose: boolean = true) {
this.getParamList().forEach(param => param.tensor.dispose()) this.getParamList().forEach(param => {
if (throwOnRedispose && param.tensor.isDisposed) {
throw new Error(`param tensor has already been disposed for path ${param.path}`)
}
param.tensor.dispose()
})
this._params = undefined this._params = undefined
} }
public async load(weightsOrUrl: Float32Array | string | undefined): Promise<void> {
if (weightsOrUrl instanceof Float32Array) {
this.extractWeights(weightsOrUrl)
return
}
if (weightsOrUrl && typeof weightsOrUrl !== 'string') {
throw new Error(`${this._name}.load - expected model uri, or weights as Float32Array`)
}
const {
paramMappings,
params
} = await this.loadQuantizedParams(weightsOrUrl)
this._paramMappings = paramMappings
this._params = params
}
public extractWeights(weights: Float32Array) {
const {
paramMappings,
params
} = this.extractParams(weights)
this._paramMappings = paramMappings
this._params = params
}
private traversePropertyPath(paramPath: string) { private traversePropertyPath(paramPath: string) {
if (!this.params) { if (!this.params) {
throw new Error(`traversePropertyPath - model has no loaded params`) throw new Error(`traversePropertyPath - model has no loaded params`)
...@@ -78,4 +113,12 @@ export class NeuralNetwork<TNetParams> { ...@@ -78,4 +113,12 @@ export class NeuralNetwork<TNetParams> {
return { obj, objProp } return { obj, objProp }
} }
protected loadQuantizedParams(_: any): Promise<{ params: TNetParams, paramMappings: ParamMapping[] }> {
throw new Error(`${this._name}.loadQuantizedParams - not implemented`)
}
protected extractParams(_: any): { params: TNetParams, paramMappings: ParamMapping[] } {
throw new Error(`${this._name}.extractParams - not implemented`)
}
} }
\ No newline at end of file
src/commons/disposeUnusedWeightTensors.ts 0 → 100644
import { ParamMapping } from './types';
export function disposeUnusedWeightTensors(weightMap: any, paramMappings: ParamMapping[]) {
Object.keys(weightMap).forEach(path => {
if (!paramMappings.some(pm => pm.originalPath === path)) {
weightMap[path].dispose()
}
})
}
src/commons/extractConvParamsFactory.ts deleted 100644 → 0
import * as tf from '@tensorflow/tfjs-core';
import { ConvParams, ExtractWeightsFunction, ParamMapping } from './types';
export function extractConvParamsFactory(extractWeights: ExtractWeightsFunction, paramMappings: ParamMapping[]) {
return function (
channelsIn: number,
channelsOut: number,
filterSize: number,
mappedPrefix: string
): ConvParams {
const filters = tf.tensor4d(
extractWeights(channelsIn * channelsOut * filterSize * filterSize),
[filterSize, filterSize, channelsIn, channelsOut]
)
const bias = tf.tensor1d(extractWeights(channelsOut))
paramMappings.push(
{ paramPath: `${mappedPrefix}/filters` },
{ paramPath: `${mappedPrefix}/bias` }
)
return {
filters,
bias
}
}
}
\ No newline at end of file
src/commons/extractWeightEntry.ts deleted 100644 → 0
import { isTensor } from './isTensor';
export function extractWeightEntry(weightMap: any, path: string, paramRank: number) {
const tensor = weightMap[path]
if (!isTensor(tensor, paramRank)) {
throw new Error(`expected weightMap[${path}] to be a Tensor${paramRank}D, instead have ${tensor}`)
}
return { path, tensor }
}
\ No newline at end of file
src/commons/extractWeightEntryFactory.ts 0 → 100644
import { isTensor } from './isTensor';
import { ParamMapping } from './types';
export function extractWeightEntryFactory(weightMap: any, paramMappings: ParamMapping[]) {
return function<T> (originalPath: string, paramRank: number, mappedPath?: string): T {
const tensor = weightMap[originalPath]
if (!isTensor(tensor, paramRank)) {
throw new Error(`expected weightMap[${originalPath}] to be a Tensor${paramRank}D, instead have ${tensor}`)
}
paramMappings.push(
{ originalPath, paramPath: mappedPath || originalPath }
)
return tensor
}
}
src/faceDetectionNet/FaceDetectionNet.ts
import * as tf from '@tensorflow/tfjs-core'; import * as tf from '@tensorflow/tfjs-core';
import { NeuralNetwork } from '../commons/NeuralNetwork';
import { NetInput } from '../NetInput'; import { NetInput } from '../NetInput';
import { Rect } from '../Rect'; import { Rect } from '../Rect';
import { toNetInput } from '../toNetInput'; import { toNetInput } from '../toNetInput';
...@@ -13,28 +14,17 @@ import { outputLayer } from './outputLayer'; ...@@ -13,28 +14,17 @@ import { outputLayer } from './outputLayer';
import { predictionLayer } from './predictionLayer'; import { predictionLayer } from './predictionLayer';
import { NetParams } from './types'; import { NetParams } from './types';
export class FaceDetectionNet { export class FaceDetectionNet extends NeuralNetwork<NetParams> {
private _params: NetParams constructor() {
super('FaceDetectionNet')
public async load(weightsOrUrl?: Float32Array | string): Promise<void> {
if (weightsOrUrl instanceof Float32Array) {
this.extractWeights(weightsOrUrl)
return
} }
if (weightsOrUrl && typeof weightsOrUrl !== 'string') { public forwardInput(input: NetInput) {
throw new Error('FaceDetectionNet.load - expected model uri, or weights as Float32Array')
}
this._params = await loadQuantizedParams(weightsOrUrl)
}
public extractWeights(weights: Float32Array) { const { params } = this
this._params = extractParams(weights)
}
public forwardInput(input: NetInput) { if (!params) {
if (!this._params) {
throw new Error('FaceDetectionNet - load model before inference') throw new Error('FaceDetectionNet - load model before inference')
} }
...@@ -42,14 +32,14 @@ export class FaceDetectionNet { ...@@ -42,14 +32,14 @@ export class FaceDetectionNet {
const batchTensor = input.toBatchTensor(512, false) const batchTensor = input.toBatchTensor(512, false)
const x = tf.sub(tf.mul(batchTensor, tf.scalar(0.007843137718737125)), tf.scalar(1)) as tf.Tensor4D const x = tf.sub(tf.mul(batchTensor, tf.scalar(0.007843137718737125)), tf.scalar(1)) as tf.Tensor4D
const features = mobileNetV1(x, this._params.mobilenetv1_params) const features = mobileNetV1(x, params.mobilenetv1)
const { const {
boxPredictions, boxPredictions,
classPredictions classPredictions
} = predictionLayer(features.out, features.conv11, this._params.prediction_layer_params) } = predictionLayer(features.out, features.conv11, params.prediction_layer)
return outputLayer(boxPredictions, classPredictions, this._params.output_layer_params) return outputLayer(boxPredictions, classPredictions, params.output_layer)
}) })
} }
...@@ -91,7 +81,6 @@ export class FaceDetectionNet { ...@@ -91,7 +81,6 @@ export class FaceDetectionNet {
minConfidence minConfidence
) )
const paddedHeightRelative = (netInput.getPaddings(0).y + netInput.getInputHeight(0)) / netInput.getInputHeight(0) const paddedHeightRelative = (netInput.getPaddings(0).y + netInput.getInputHeight(0)) / netInput.getInputHeight(0)
const paddedWidthRelative = (netInput.getPaddings(0).x + netInput.getInputWidth(0)) / netInput.getInputWidth(0) const paddedWidthRelative = (netInput.getPaddings(0).x + netInput.getInputWidth(0)) / netInput.getInputWidth(0)
...@@ -125,4 +114,12 @@ export class FaceDetectionNet { ...@@ -125,4 +114,12 @@ export class FaceDetectionNet {
return results return results
} }
protected loadQuantizedParams(uri: string | undefined) {
return loadQuantizedParams(uri)
}
protected extractParams(weights: Float32Array) {
return extractParams(weights)
}
} }
\ No newline at end of file
src/faceDetectionNet/boxPredictionLayer.ts
...@@ -13,11 +13,11 @@ export function boxPredictionLayer( ...@@ -13,11 +13,11 @@ export function boxPredictionLayer(
const batchSize = x.shape[0] const batchSize = x.shape[0]
const boxPredictionEncoding = tf.reshape( const boxPredictionEncoding = tf.reshape(
convLayer(x, params.box_encoding_predictor_params), convLayer(x, params.box_encoding_predictor),
[batchSize, -1, 1, 4] [batchSize, -1, 1, 4]
) )
const classPrediction = tf.reshape( const classPrediction = tf.reshape(
convLayer(x, params.class_predictor_params), convLayer(x, params.class_predictor),
[batchSize, -1, 3] [batchSize, -1, 3]
) )
......
src/faceDetectionNet/extractParams.ts
import * as tf from '@tensorflow/tfjs-core'; import * as tf from '@tensorflow/tfjs-core';
import { extractWeightsFactory } from '../commons/extractWeightsFactory'; import { extractWeightsFactory } from '../commons/extractWeightsFactory';
import { ConvParams } from '../commons/types'; import { ConvParams, ExtractWeightsFunction, ParamMapping } from '../commons/types';
import { MobileNetV1, NetParams, PointwiseConvParams, PredictionLayerParams } from './types'; import { MobileNetV1, NetParams, PointwiseConvParams, PredictionLayerParams } from './types';
function extractorsFactory(extractWeights: (numWeights: number) => Float32Array) { function extractorsFactory(extractWeights: ExtractWeightsFunction, paramMappings: ParamMapping[]) {
function extractDepthwiseConvParams(numChannels: number, mappedPrefix: string): MobileNetV1.DepthwiseConvParams {
function extractDepthwiseConvParams(numChannels: number): MobileNetV1.DepthwiseConvParams {
const filters = tf.tensor4d(extractWeights(3 * 3 * numChannels), [3, 3, numChannels, 1]) const filters = tf.tensor4d(extractWeights(3 * 3 * numChannels), [3, 3, numChannels, 1])
const batch_norm_scale = tf.tensor1d(extractWeights(numChannels)) const batch_norm_scale = tf.tensor1d(extractWeights(numChannels))
const batch_norm_offset = tf.tensor1d(extractWeights(numChannels)) const batch_norm_offset = tf.tensor1d(extractWeights(numChannels))
const batch_norm_mean = tf.tensor1d(extractWeights(numChannels)) const batch_norm_mean = tf.tensor1d(extractWeights(numChannels))
const batch_norm_variance = tf.tensor1d(extractWeights(numChannels)) const batch_norm_variance = tf.tensor1d(extractWeights(numChannels))
paramMappings.push(
{ paramPath: `${mappedPrefix}/filters` },
{ paramPath: `${mappedPrefix}/batch_norm_scale` },
{ paramPath: `${mappedPrefix}/batch_norm_offset` },
{ paramPath: `${mappedPrefix}/batch_norm_mean` },
{ paramPath: `${mappedPrefix}/batch_norm_variance` }
)
return { return {
filters, filters,
batch_norm_scale, batch_norm_scale,
...@@ -25,29 +34,36 @@ function extractorsFactory(extractWeights: (numWeights: number) => Float32Array) ...@@ -25,29 +34,36 @@ function extractorsFactory(extractWeights: (numWeights: number) => Float32Array)
function extractConvParams( function extractConvParams(
channelsIn: number, channelsIn: number,
channelsOut: number, channelsOut: number,
filterSize: number filterSize: number,
mappedPrefix: string,
isPointwiseConv?: boolean
): ConvParams { ): ConvParams {
const filters = tf.tensor4d( const filters = tf.tensor4d(
extractWeights(channelsIn * channelsOut * filterSize * filterSize), extractWeights(channelsIn * channelsOut * filterSize * filterSize),
[filterSize, filterSize, channelsIn, channelsOut] [filterSize, filterSize, channelsIn, channelsOut]
) )
const bias = tf.tensor1d(extractWeights(channelsOut)) const bias = tf.tensor1d(extractWeights(channelsOut))
return { paramMappings.push(
filters, { paramPath: `${mappedPrefix}/filters` },
bias { paramPath: `${mappedPrefix}/${isPointwiseConv ? 'batch_norm_offset' : 'bias'}` }
} )
return { filters, bias }
} }
function extractPointwiseConvParams( function extractPointwiseConvParams(
channelsIn: number, channelsIn: number,
channelsOut: number, channelsOut: number,
filterSize: number filterSize: number,
mappedPrefix: string
): PointwiseConvParams { ): PointwiseConvParams {
const { const {
filters, filters,
bias bias
} = extractConvParams(channelsIn, channelsOut, filterSize) } = extractConvParams(channelsIn, channelsOut, filterSize, mappedPrefix, true)
return { return {
filters, filters,
...@@ -57,115 +73,118 @@ function extractorsFactory(extractWeights: (numWeights: number) => Float32Array) ...@@ -57,115 +73,118 @@ function extractorsFactory(extractWeights: (numWeights: number) => Float32Array)
function extractConvPairParams( function extractConvPairParams(
channelsIn: number, channelsIn: number,
channelsOut: number channelsOut: number,
mappedPrefix: string
): MobileNetV1.ConvPairParams { ): MobileNetV1.ConvPairParams {
const depthwise_conv_params = extractDepthwiseConvParams(channelsIn)
const pointwise_conv_params = extractPointwiseConvParams(channelsIn, channelsOut, 1)
return { const depthwise_conv = extractDepthwiseConvParams(channelsIn, `${mappedPrefix}/depthwise_conv`)
depthwise_conv_params, const pointwise_conv = extractPointwiseConvParams(channelsIn, channelsOut, 1, `${mappedPrefix}/pointwise_conv`)
pointwise_conv_params
} return { depthwise_conv, pointwise_conv }
} }
function extractMobilenetV1Params(): MobileNetV1.Params { function extractMobilenetV1Params(): MobileNetV1.Params {
const conv_0_params = extractPointwiseConvParams(3, 32, 3) const conv_0 = extractPointwiseConvParams(3, 32, 3, 'mobilenetv1/conv_0')
const channelNumPairs = [ const conv_1 = extractConvPairParams(32, 64, 'mobilenetv1/conv_1')
[32, 64], const conv_2 = extractConvPairParams(64, 128, 'mobilenetv1/conv_2')
[64, 128], const conv_3 = extractConvPairParams(128, 128, 'mobilenetv1/conv_3')
[128, 128], const conv_4 = extractConvPairParams(128, 256, 'mobilenetv1/conv_4')
[128, 256], const conv_5 = extractConvPairParams(256, 256, 'mobilenetv1/conv_5')
[256, 256], const conv_6 = extractConvPairParams(256, 512, 'mobilenetv1/conv_6')
[256, 512], const conv_7 = extractConvPairParams(512, 512, 'mobilenetv1/conv_7')
[512, 512], const conv_8 = extractConvPairParams(512, 512, 'mobilenetv1/conv_8')
[512, 512], const conv_9 = extractConvPairParams(512, 512, 'mobilenetv1/conv_9')
[512, 512], const conv_10 = extractConvPairParams(512, 512, 'mobilenetv1/conv_10')
[512, 512], const conv_11 = extractConvPairParams(512, 512, 'mobilenetv1/conv_11')
[512, 512], const conv_12 = extractConvPairParams(512, 1024, 'mobilenetv1/conv_12')
[512, 1024], const conv_13 = extractConvPairParams(1024, 1024, 'mobilenetv1/conv_13')
[1024, 1024]
]
const conv_pair_params = channelNumPairs.map(
([channelsIn, channelsOut]) => extractConvPairParams(channelsIn, channelsOut)
)
return { return {
conv_0_params, conv_0,
conv_pair_params conv_1,
conv_2,
conv_3,
conv_4,
conv_5,
conv_6,
conv_7,
conv_8,
conv_9,
conv_10,
conv_11,
conv_12,
conv_13
} }
} }
function extractPredictionLayerParams(): PredictionLayerParams { function extractPredictionLayerParams(): PredictionLayerParams {
const conv_0_params = extractPointwiseConvParams(1024, 256, 1) const conv_0 = extractPointwiseConvParams(1024, 256, 1, 'prediction_layer/conv_0')
const conv_1_params = extractPointwiseConvParams(256, 512, 3) const conv_1 = extractPointwiseConvParams(256, 512, 3, 'prediction_layer/conv_1')
const conv_2_params = extractPointwiseConvParams(512, 128, 1) const conv_2 = extractPointwiseConvParams(512, 128, 1, 'prediction_layer/conv_2')
const conv_3_params = extractPointwiseConvParams(128, 256, 3) const conv_3 = extractPointwiseConvParams(128, 256, 3, 'prediction_layer/conv_3')
const conv_4_params = extractPointwiseConvParams(256, 128, 1) const conv_4 = extractPointwiseConvParams(256, 128, 1, 'prediction_layer/conv_4')
const conv_5_params = extractPointwiseConvParams(128, 256, 3) const conv_5 = extractPointwiseConvParams(128, 256, 3, 'prediction_layer/conv_5')
const conv_6_params = extractPointwiseConvParams(256, 64, 1) const conv_6 = extractPointwiseConvParams(256, 64, 1, 'prediction_layer/conv_6')
const conv_7_params = extractPointwiseConvParams(64, 128, 3) const conv_7 = extractPointwiseConvParams(64, 128, 3, 'prediction_layer/conv_7')
const box_encoding_0_predictor_params = extractConvParams(512, 12, 1) const box_encoding_0_predictor = extractConvParams(512, 12, 1, 'prediction_layer/box_predictor_0/box_encoding_predictor')
const class_predictor_0_params = extractConvParams(512, 9, 1) const class_predictor_0 = extractConvParams(512, 9, 1, 'prediction_layer/box_predictor_0/class_predictor')
const box_encoding_1_predictor_params = extractConvParams(1024, 24, 1) const box_encoding_1_predictor = extractConvParams(1024, 24, 1, 'prediction_layer/box_predictor_1/box_encoding_predictor')
const class_predictor_1_params = extractConvParams(1024, 18, 1) const class_predictor_1 = extractConvParams(1024, 18, 1, 'prediction_layer/box_predictor_1/class_predictor')
const box_encoding_2_predictor_params = extractConvParams(512, 24, 1) const box_encoding_2_predictor = extractConvParams(512, 24, 1, 'prediction_layer/box_predictor_2/box_encoding_predictor')
const class_predictor_2_params = extractConvParams(512, 18, 1) const class_predictor_2 = extractConvParams(512, 18, 1, 'prediction_layer/box_predictor_2/class_predictor')
const box_encoding_3_predictor_params = extractConvParams(256, 24, 1) const box_encoding_3_predictor = extractConvParams(256, 24, 1, 'prediction_layer/box_predictor_3/box_encoding_predictor')
const class_predictor_3_params = extractConvParams(256, 18, 1) const class_predictor_3 = extractConvParams(256, 18, 1, 'prediction_layer/box_predictor_3/class_predictor')
const box_encoding_4_predictor_params = extractConvParams(256, 24, 1) const box_encoding_4_predictor = extractConvParams(256, 24, 1, 'prediction_layer/box_predictor_4/box_encoding_predictor')
const class_predictor_4_params = extractConvParams(256, 18, 1) const class_predictor_4 = extractConvParams(256, 18, 1, 'prediction_layer/box_predictor_4/class_predictor')
const box_encoding_5_predictor_params = extractConvParams(128, 24, 1) const box_encoding_5_predictor = extractConvParams(128, 24, 1, 'prediction_layer/box_predictor_5/box_encoding_predictor')
const class_predictor_5_params = extractConvParams(128, 18, 1) const class_predictor_5 = extractConvParams(128, 18, 1, 'prediction_layer/box_predictor_5/class_predictor')
const box_predictor_0_params = { const box_predictor_0 = {
box_encoding_predictor_params: box_encoding_0_predictor_params, box_encoding_predictor: box_encoding_0_predictor,
class_predictor_params: class_predictor_0_params class_predictor: class_predictor_0
} }
const box_predictor_1_params = { const box_predictor_1 = {
box_encoding_predictor_params: box_encoding_1_predictor_params, box_encoding_predictor: box_encoding_1_predictor,
class_predictor_params: class_predictor_1_params class_predictor: class_predictor_1
} }
const box_predictor_2_params = { const box_predictor_2 = {
box_encoding_predictor_params: box_encoding_2_predictor_params, box_encoding_predictor: box_encoding_2_predictor,
class_predictor_params: class_predictor_2_params class_predictor: class_predictor_2
} }
const box_predictor_3_params = { const box_predictor_3 = {
box_encoding_predictor_params: box_encoding_3_predictor_params, box_encoding_predictor: box_encoding_3_predictor,
class_predictor_params: class_predictor_3_params class_predictor: class_predictor_3
} }
const box_predictor_4_params = { const box_predictor_4 = {
box_encoding_predictor_params: box_encoding_4_predictor_params, box_encoding_predictor: box_encoding_4_predictor,
class_predictor_params: class_predictor_4_params class_predictor: class_predictor_4
} }
const box_predictor_5_params = { const box_predictor_5 = {
box_encoding_predictor_params: box_encoding_5_predictor_params, box_encoding_predictor: box_encoding_5_predictor,
class_predictor_params: class_predictor_5_params class_predictor: class_predictor_5
} }
return { return {
conv_0_params, conv_0,
conv_1_params, conv_1,
conv_2_params, conv_2,
conv_3_params, conv_3,
conv_4_params, conv_4,
conv_5_params, conv_5,
conv_6_params, conv_6,
conv_7_params, conv_7,
box_predictor_0_params, box_predictor_0,
box_predictor_1_params, box_predictor_1,
box_predictor_2_params, box_predictor_2,
box_predictor_3_params, box_predictor_3,
box_predictor_4_params, box_predictor_4,
box_predictor_5_params box_predictor_5
} }
} }
return { return {
extractMobilenetV1Params, extractMobilenetV1Params,
extractPredictionLayerParams extractPredictionLayerParams
...@@ -173,7 +192,10 @@ function extractorsFactory(extractWeights: (numWeights: number) => Float32Array) ...@@ -173,7 +192,10 @@ function extractorsFactory(extractWeights: (numWeights: number) => Float32Array)
} }
export function extractParams(weights: Float32Array): NetParams { export function extractParams(weights: Float32Array): { params: NetParams, paramMappings: ParamMapping[] } {
const paramMappings: ParamMapping[] = []
const { const {
extractWeights, extractWeights,
getRemainingWeights getRemainingWeights
...@@ -182,25 +204,30 @@ export function extractParams(weights: Float32Array): NetParams { ...@@ -182,25 +204,30 @@ export function extractParams(weights: Float32Array): NetParams {
const { const {
extractMobilenetV1Params, extractMobilenetV1Params,
extractPredictionLayerParams extractPredictionLayerParams
} = extractorsFactory(extractWeights) } = extractorsFactory(extractWeights, paramMappings)
const mobilenetv1_params = extractMobilenetV1Params() const mobilenetv1 = extractMobilenetV1Params()
const prediction_layer_params = extractPredictionLayerParams() const prediction_layer = extractPredictionLayerParams()
const extra_dim = tf.tensor3d( const extra_dim = tf.tensor3d(
extractWeights(5118 * 4), extractWeights(5118 * 4),
[1, 5118, 4] [1, 5118, 4]
) )
const output_layer_params = { const output_layer = {
extra_dim extra_dim
} }
paramMappings.push({ paramPath: 'output_layer/extra_dim' })
if (getRemainingWeights().length !== 0) { if (getRemainingWeights().length !== 0) {
throw new Error(`weights remaing after extract: ${getRemainingWeights().length}`) throw new Error(`weights remaing after extract: ${getRemainingWeights().length}`)
} }
return { return {
mobilenetv1_params, params: {
prediction_layer_params, mobilenetv1,
output_layer_params prediction_layer,
output_layer
},
paramMappings
} }
} }
\ No newline at end of file
src/faceDetectionNet/loadQuantizedParams.ts
import { isTensor1D, isTensor4D, isTensor3D } from '../commons/isTensor'; import { tf } from '..';
import { disposeUnusedWeightTensors } from '../commons/disposeUnusedWeightTensors';
import { extractWeightEntryFactory } from '../commons/extractWeightEntryFactory';
import { isTensor1D, isTensor3D, isTensor4D } from '../commons/isTensor';
import { loadWeightMap } from '../commons/loadWeightMap'; import { loadWeightMap } from '../commons/loadWeightMap';
import { BoxPredictionParams, MobileNetV1, PointwiseConvParams, PredictionLayerParams } from './types'; import { ConvParams, ParamMapping } from '../commons/types';
import { BoxPredictionParams, MobileNetV1, NetParams, PointwiseConvParams, PredictionLayerParams } from './types';
const DEFAULT_MODEL_NAME = 'face_detection_model' const DEFAULT_MODEL_NAME = 'face_detection_model'
function extractorsFactory(weightMap: any) { function extractorsFactory(weightMap: any, paramMappings: ParamMapping[]) {
function extractPointwiseConvParams(prefix: string, idx: number): PointwiseConvParams { const extractWeightEntry = extractWeightEntryFactory(weightMap, paramMappings)
const pointwise_conv_params = { function extractPointwiseConvParams(prefix: string, idx: number, mappedPrefix: string): PointwiseConvParams {
filters: weightMap[`${prefix}/Conv2d_${idx}_pointwise/weights`],
batch_norm_offset: weightMap[`${prefix}/Conv2d_${idx}_pointwise/convolution_bn_offset`]
}
if (!isTensor4D(pointwise_conv_params.filters)) {
throw new Error(`expected weightMap[${prefix}/Conv2d_${idx}_pointwise/weights] to be a Tensor4D, instead have ${pointwise_conv_params.filters}`)
}
if (!isTensor1D(pointwise_conv_params.batch_norm_offset)) { const filters = extractWeightEntry<tf.Tensor4D>(`${prefix}/Conv2d_${idx}_pointwise/weights`, 4, `${mappedPrefix}/filters`)
throw new Error(`expected weightMap[${prefix}/Conv2d_${idx}_pointwise/convolution_bn_offset] to be a Tensor1D, instead have ${pointwise_conv_params.batch_norm_offset}`) const batch_norm_offset = extractWeightEntry<tf.Tensor1D>(`${prefix}/Conv2d_${idx}_pointwise/convolution_bn_offset`, 1, `${mappedPrefix}/batch_norm_offset`)
}
return pointwise_conv_params return { filters, batch_norm_offset }
} }
function extractConvPairParams(idx: number): MobileNetV1.ConvPairParams { function extractConvPairParams(idx: number): MobileNetV1.ConvPairParams {
const depthwise_conv_params = { const mappedPrefix = `mobilenetv1/conv_${idx}`
filters: weightMap[`MobilenetV1/Conv2d_${idx}_depthwise/depthwise_weights`], const prefixDepthwiseConv = `MobilenetV1/Conv2d_${idx}_depthwise`
batch_norm_scale: weightMap[`MobilenetV1/Conv2d_${idx}_depthwise/BatchNorm/gamma`], const mappedPrefixDepthwiseConv = `${mappedPrefix}/depthwise_conv`
batch_norm_offset: weightMap[`MobilenetV1/Conv2d_${idx}_depthwise/BatchNorm/beta`], const mappedPrefixPointwiseConv = `${mappedPrefix}/pointwise_conv`
batch_norm_mean: weightMap[`MobilenetV1/Conv2d_${idx}_depthwise/BatchNorm/moving_mean`],
batch_norm_variance: weightMap[`MobilenetV1/Conv2d_${idx}_depthwise/BatchNorm/moving_variance`],
}
if (!isTensor4D(depthwise_conv_params.filters)) {
throw new Error(`expected weightMap[MobilenetV1/Conv2d_${idx}_depthwise/depthwise_weights] to be a Tensor4D, instead have ${depthwise_conv_params.filters}`)
}
if (!isTensor1D(depthwise_conv_params.batch_norm_scale)) {
throw new Error(`expected weightMap[MobilenetV1/Conv2d_${idx}_depthwise/BatchNorm/gamma] to be a Tensor1D, instead have ${depthwise_conv_params.batch_norm_scale}`)
}
if (!isTensor1D(depthwise_conv_params.batch_norm_offset)) {
throw new Error(`expected weightMap[MobilenetV1/Conv2d_${idx}_depthwise/BatchNorm/beta] to be a Tensor1D, instead have ${depthwise_conv_params.batch_norm_offset}`)
}
if (!isTensor1D(depthwise_conv_params.batch_norm_mean)) {
throw new Error(`expected weightMap[MobilenetV1/Conv2d_${idx}_depthwise/BatchNorm/moving_mean] to be a Tensor1D, instead have ${depthwise_conv_params.batch_norm_mean}`)
}
if (!isTensor1D(depthwise_conv_params.batch_norm_variance)) { const filters = extractWeightEntry<tf.Tensor4D>(`${prefixDepthwiseConv}/depthwise_weights`, 4, `${mappedPrefixDepthwiseConv}/filters`)
throw new Error(`expected weightMap[MobilenetV1/Conv2d_${idx}_depthwise/BatchNorm/moving_variance] to be a Tensor1D, instead have ${depthwise_conv_params.batch_norm_variance}`) const batch_norm_scale = extractWeightEntry<tf.Tensor1D>(`${prefixDepthwiseConv}/BatchNorm/gamma`, 1, `${mappedPrefixDepthwiseConv}/batch_norm_scale`)
} const batch_norm_offset = extractWeightEntry<tf.Tensor1D>(`${prefixDepthwiseConv}/BatchNorm/beta`, 1, `${mappedPrefixDepthwiseConv}/batch_norm_offset`)
const batch_norm_mean = extractWeightEntry<tf.Tensor1D>(`${prefixDepthwiseConv}/BatchNorm/moving_mean`, 1, `${mappedPrefixDepthwiseConv}/batch_norm_mean`)
const batch_norm_variance = extractWeightEntry<tf.Tensor1D>(`${prefixDepthwiseConv}/BatchNorm/moving_variance`, 1, `${mappedPrefixDepthwiseConv}/batch_norm_variance`)
return { return {
depthwise_conv_params, depthwise_conv: {
pointwise_conv_params: extractPointwiseConvParams('MobilenetV1', idx) filters,
batch_norm_scale,
batch_norm_offset,
batch_norm_mean,
batch_norm_variance
},
pointwise_conv: extractPointwiseConvParams('MobilenetV1', idx, mappedPrefixPointwiseConv)
} }
} }
function extractMobilenetV1Params(): MobileNetV1.Params { function extractMobilenetV1Params(): MobileNetV1.Params {
return { return {
conv_0_params: extractPointwiseConvParams('MobilenetV1', 0), conv_0: extractPointwiseConvParams('MobilenetV1', 0, 'mobilenetv1/conv_0'),
conv_pair_params: Array(13).fill(0).map((_, i) => extractConvPairParams(i + 1)) conv_1: extractConvPairParams(1),
} conv_2: extractConvPairParams(2),
} conv_3: extractConvPairParams(3),
conv_4: extractConvPairParams(4),
function extractBoxPredictorParams(idx: number): BoxPredictionParams { conv_5: extractConvPairParams(5),
conv_6: extractConvPairParams(6),
const params = { conv_7: extractConvPairParams(7),
box_encoding_predictor_params: { conv_8: extractConvPairParams(8),
filters: weightMap[`Prediction/BoxPredictor_${idx}/BoxEncodingPredictor/weights`], conv_9: extractConvPairParams(9),
bias: weightMap[`Prediction/BoxPredictor_${idx}/BoxEncodingPredictor/biases`] conv_10: extractConvPairParams(10),
}, conv_11: extractConvPairParams(11),
class_predictor_params: { conv_12: extractConvPairParams(12),
filters: weightMap[`Prediction/BoxPredictor_${idx}/ClassPredictor/weights`], conv_13: extractConvPairParams(13)
bias: weightMap[`Prediction/BoxPredictor_${idx}/ClassPredictor/biases`]
} }
} }
if (!isTensor4D(params.box_encoding_predictor_params.filters)) { function extractConvParams(prefix: string, mappedPrefix: string): ConvParams {
throw new Error(`expected weightMap[Prediction/BoxPredictor_${idx}/BoxEncodingPredictor/weights] to be a Tensor4D, instead have ${params.box_encoding_predictor_params.filters}`) const filters = extractWeightEntry<tf.Tensor4D>(`${prefix}/weights`, 4, `${mappedPrefix}/filters`)
} const bias = extractWeightEntry<tf.Tensor1D>(`${prefix}/biases`, 1, `${mappedPrefix}/bias`)
if (!isTensor1D(params.box_encoding_predictor_params.bias)) { return { filters, bias }
throw new Error(`expected weightMap[Prediction/BoxPredictor_${idx}/BoxEncodingPredictor/biases] to be a Tensor1D, instead have ${params.box_encoding_predictor_params.bias}`)
} }
if (!isTensor4D(params.class_predictor_params.filters)) { function extractBoxPredictorParams(idx: number): BoxPredictionParams {
throw new Error(`expected weightMap[Prediction/BoxPredictor_${idx}/ClassPredictor/weights] to be a Tensor4D, instead have ${params.class_predictor_params.filters}`)
}
if (!isTensor1D(params.class_predictor_params.bias)) { const box_encoding_predictor = extractConvParams(
throw new Error(`expected weightMap[Prediction/BoxPredictor_${idx}/ClassPredictor/biases] to be a Tensor1D, instead have ${params.class_predictor_params.bias}`) `Prediction/BoxPredictor_${idx}/BoxEncodingPredictor`,
} `prediction_layer/box_predictor_${idx}/box_encoding_predictor`
)
const class_predictor = extractConvParams(
`Prediction/BoxPredictor_${idx}/ClassPredictor`,
`prediction_layer/box_predictor_${idx}/class_predictor`
)
return params return { box_encoding_predictor, class_predictor }
} }
function extractPredictionLayerParams(): PredictionLayerParams { function extractPredictionLayerParams(): PredictionLayerParams {
return { return {
conv_0_params: extractPointwiseConvParams('Prediction', 0), conv_0: extractPointwiseConvParams('Prediction', 0, 'prediction_layer/conv_0'),
conv_1_params: extractPointwiseConvParams('Prediction', 1), conv_1: extractPointwiseConvParams('Prediction', 1, 'prediction_layer/conv_1'),
conv_2_params: extractPointwiseConvParams('Prediction', 2), conv_2: extractPointwiseConvParams('Prediction', 2, 'prediction_layer/conv_2'),
conv_3_params: extractPointwiseConvParams('Prediction', 3), conv_3: extractPointwiseConvParams('Prediction', 3, 'prediction_layer/conv_3'),
conv_4_params: extractPointwiseConvParams('Prediction', 4), conv_4: extractPointwiseConvParams('Prediction', 4, 'prediction_layer/conv_4'),
conv_5_params: extractPointwiseConvParams('Prediction', 5), conv_5: extractPointwiseConvParams('Prediction', 5, 'prediction_layer/conv_5'),
conv_6_params: extractPointwiseConvParams('Prediction', 6), conv_6: extractPointwiseConvParams('Prediction', 6, 'prediction_layer/conv_6'),
conv_7_params: extractPointwiseConvParams('Prediction', 7), conv_7: extractPointwiseConvParams('Prediction', 7, 'prediction_layer/conv_7'),
box_predictor_0_params: extractBoxPredictorParams(0), box_predictor_0: extractBoxPredictorParams(0),
box_predictor_1_params: extractBoxPredictorParams(1), box_predictor_1: extractBoxPredictorParams(1),
box_predictor_2_params: extractBoxPredictorParams(2), box_predictor_2: extractBoxPredictorParams(2),
box_predictor_3_params: extractBoxPredictorParams(3), box_predictor_3: extractBoxPredictorParams(3),
box_predictor_4_params: extractBoxPredictorParams(4), box_predictor_4: extractBoxPredictorParams(4),
box_predictor_5_params: extractBoxPredictorParams(5) box_predictor_5: extractBoxPredictorParams(5)
} }
} }
...@@ -124,24 +110,34 @@ function extractorsFactory(weightMap: any) { ...@@ -124,24 +110,34 @@ function extractorsFactory(weightMap: any) {
} }
} }
export async function loadQuantizedParams(uri: string | undefined): Promise<any> {//Promise<NetParams> { export async function loadQuantizedParams(
uri: string | undefined
): Promise<{ params: NetParams, paramMappings: ParamMapping[] }> {
const weightMap = await loadWeightMap(uri, DEFAULT_MODEL_NAME) const weightMap = await loadWeightMap(uri, DEFAULT_MODEL_NAME)
const paramMappings: ParamMapping[] = []
const { const {
extractMobilenetV1Params, extractMobilenetV1Params,
extractPredictionLayerParams extractPredictionLayerParams
} = extractorsFactory(weightMap) } = extractorsFactory(weightMap, paramMappings)
const extra_dim = weightMap['Output/extra_dim'] const extra_dim = weightMap['Output/extra_dim']
paramMappings.push({ originalPath: 'Output/extra_dim', paramPath: 'output_layer/extra_dim' })
if (!isTensor3D(extra_dim)) { if (!isTensor3D(extra_dim)) {
throw new Error(`expected weightMap['Output/extra_dim'] to be a Tensor3D, instead have ${extra_dim}`) throw new Error(`expected weightMap['Output/extra_dim'] to be a Tensor3D, instead have ${extra_dim}`)
} }
return { const params = {
mobilenetv1_params: extractMobilenetV1Params(), mobilenetv1: extractMobilenetV1Params(),
prediction_layer_params: extractPredictionLayerParams(), prediction_layer: extractPredictionLayerParams(),
output_layer_params: { output_layer: {
extra_dim extra_dim
} }
} }
disposeUnusedWeightTensors(weightMap, paramMappings)
return { params, paramMappings }
} }
\ No newline at end of file
src/faceDetectionNet/mobileNetV1.ts
...@@ -34,13 +34,29 @@ export function mobileNetV1(x: tf.Tensor4D, params: MobileNetV1.Params) { ...@@ -34,13 +34,29 @@ export function mobileNetV1(x: tf.Tensor4D, params: MobileNetV1.Params) {
return tf.tidy(() => { return tf.tidy(() => {
let conv11 = null let conv11 = null
let out = pointwiseConvLayer(x, params.conv_0_params, [2, 2]) let out = pointwiseConvLayer(x, params.conv_0, [2, 2])
params.conv_pair_params.forEach((param, i) => { const convPairParams = [
params.conv_1,
params.conv_2,
params.conv_3,
params.conv_4,
params.conv_5,
params.conv_6,
params.conv_7,
params.conv_8,
params.conv_9,
params.conv_10,
params.conv_11,
params.conv_12,
params.conv_13
]
convPairParams.forEach((param, i) => {
const layerIdx = i + 1 const layerIdx = i + 1
const depthwiseConvStrides = getStridesForLayerIdx(layerIdx) const depthwiseConvStrides = getStridesForLayerIdx(layerIdx)
out = depthwiseConvLayer(out, param.depthwise_conv_params, depthwiseConvStrides) out = depthwiseConvLayer(out, param.depthwise_conv, depthwiseConvStrides)
out = pointwiseConvLayer(out, param.pointwise_conv_params, [1, 1]) out = pointwiseConvLayer(out, param.pointwise_conv, [1, 1])
if (layerIdx === 11) { if (layerIdx === 11) {
conv11 = out conv11 = out
} }
......
src/faceDetectionNet/predictionLayer.ts
...@@ -11,21 +11,21 @@ export function predictionLayer( ...@@ -11,21 +11,21 @@ export function predictionLayer(
) { ) {
return tf.tidy(() => { return tf.tidy(() => {
const conv0 = pointwiseConvLayer(x, params.conv_0_params, [1, 1]) const conv0 = pointwiseConvLayer(x, params.conv_0, [1, 1])
const conv1 = pointwiseConvLayer(conv0, params.conv_1_params, [2, 2]) const conv1 = pointwiseConvLayer(conv0, params.conv_1, [2, 2])
const conv2 = pointwiseConvLayer(conv1, params.conv_2_params, [1, 1]) const conv2 = pointwiseConvLayer(conv1, params.conv_2, [1, 1])
const conv3 = pointwiseConvLayer(conv2, params.conv_3_params, [2, 2]) const conv3 = pointwiseConvLayer(conv2, params.conv_3, [2, 2])
const conv4 = pointwiseConvLayer(conv3, params.conv_4_params, [1, 1]) const conv4 = pointwiseConvLayer(conv3, params.conv_4, [1, 1])
const conv5 = pointwiseConvLayer(conv4, params.conv_5_params, [2, 2]) const conv5 = pointwiseConvLayer(conv4, params.conv_5, [2, 2])
const conv6 = pointwiseConvLayer(conv5, params.conv_6_params, [1, 1]) const conv6 = pointwiseConvLayer(conv5, params.conv_6, [1, 1])
const conv7 = pointwiseConvLayer(conv6, params.conv_7_params, [2, 2]) const conv7 = pointwiseConvLayer(conv6, params.conv_7, [2, 2])
const boxPrediction0 = boxPredictionLayer(conv11, params.box_predictor_0_params) const boxPrediction0 = boxPredictionLayer(conv11, params.box_predictor_0)
const boxPrediction1 = boxPredictionLayer(x, params.box_predictor_1_params) const boxPrediction1 = boxPredictionLayer(x, params.box_predictor_1)
const boxPrediction2 = boxPredictionLayer(conv1, params.box_predictor_2_params) const boxPrediction2 = boxPredictionLayer(conv1, params.box_predictor_2)
const boxPrediction3 = boxPredictionLayer(conv3, params.box_predictor_3_params) const boxPrediction3 = boxPredictionLayer(conv3, params.box_predictor_3)
const boxPrediction4 = boxPredictionLayer(conv5, params.box_predictor_4_params) const boxPrediction4 = boxPredictionLayer(conv5, params.box_predictor_4)
const boxPrediction5 = boxPredictionLayer(conv7, params.box_predictor_5_params) const boxPrediction5 = boxPredictionLayer(conv7, params.box_predictor_5)
const boxPredictions = tf.concat([ const boxPredictions = tf.concat([
boxPrediction0.boxPredictionEncoding, boxPrediction0.boxPredictionEncoding,
......
src/faceDetectionNet/types.ts
...@@ -18,37 +18,49 @@ export namespace MobileNetV1 { ...@@ -18,37 +18,49 @@ export namespace MobileNetV1 {
} }
export type ConvPairParams = { export type ConvPairParams = {
depthwise_conv_params: DepthwiseConvParams depthwise_conv: DepthwiseConvParams
pointwise_conv_params: PointwiseConvParams pointwise_conv: PointwiseConvParams
} }
export type Params = { export type Params = {
conv_0_params: PointwiseConvParams conv_0: PointwiseConvParams
conv_pair_params: ConvPairParams[] conv_1: ConvPairParams
conv_2: ConvPairParams
conv_3: ConvPairParams
conv_4: ConvPairParams
conv_5: ConvPairParams
conv_6: ConvPairParams
conv_7: ConvPairParams
conv_8: ConvPairParams
conv_9: ConvPairParams
conv_10: ConvPairParams
conv_11: ConvPairParams
conv_12: ConvPairParams
conv_13: ConvPairParams
} }
} }
export type BoxPredictionParams = { export type BoxPredictionParams = {
box_encoding_predictor_params: ConvParams box_encoding_predictor: ConvParams
class_predictor_params: ConvParams class_predictor: ConvParams
} }
export type PredictionLayerParams = { export type PredictionLayerParams = {
conv_0_params: PointwiseConvParams conv_0: PointwiseConvParams
conv_1_params: PointwiseConvParams conv_1: PointwiseConvParams
conv_2_params: PointwiseConvParams conv_2: PointwiseConvParams
conv_3_params: PointwiseConvParams conv_3: PointwiseConvParams
conv_4_params: PointwiseConvParams conv_4: PointwiseConvParams
conv_5_params: PointwiseConvParams conv_5: PointwiseConvParams
conv_6_params: PointwiseConvParams conv_6: PointwiseConvParams
conv_7_params: PointwiseConvParams conv_7: PointwiseConvParams
box_predictor_0_params: BoxPredictionParams box_predictor_0: BoxPredictionParams
box_predictor_1_params: BoxPredictionParams box_predictor_1: BoxPredictionParams
box_predictor_2_params: BoxPredictionParams box_predictor_2: BoxPredictionParams
box_predictor_3_params: BoxPredictionParams box_predictor_3: BoxPredictionParams
box_predictor_4_params: BoxPredictionParams box_predictor_4: BoxPredictionParams
box_predictor_5_params: BoxPredictionParams box_predictor_5: BoxPredictionParams
} }
export type OutputLayerParams = { export type OutputLayerParams = {
...@@ -56,7 +68,7 @@ export type OutputLayerParams = { ...@@ -56,7 +68,7 @@ export type OutputLayerParams = {
} }
export type NetParams = { export type NetParams = {
mobilenetv1_params: MobileNetV1.Params, mobilenetv1: MobileNetV1.Params,
prediction_layer_params: PredictionLayerParams, prediction_layer: PredictionLayerParams,
output_layer_params: OutputLayerParams output_layer: OutputLayerParams
} }
src/faceLandmarkNet/FaceLandmarkNet.ts
...@@ -24,36 +24,13 @@ function maxPool(x: tf.Tensor4D, strides: [number, number] = [2, 2]): tf.Tensor4 ...@@ -24,36 +24,13 @@ function maxPool(x: tf.Tensor4D, strides: [number, number] = [2, 2]): tf.Tensor4
export class FaceLandmarkNet extends NeuralNetwork<NetParams> { export class FaceLandmarkNet extends NeuralNetwork<NetParams> {
public async load(weightsOrUrl: Float32Array | string | undefined): Promise<void> { constructor() {
if (weightsOrUrl instanceof Float32Array) { super('FaceLandmarkNet')
this.extractWeights(weightsOrUrl)
return
}
if (weightsOrUrl && typeof weightsOrUrl !== 'string') {
throw new Error('FaceLandmarkNet.load - expected model uri, or weights as Float32Array')
}
const {
paramMappings,
params
} = await loadQuantizedParams(weightsOrUrl)
this._paramMappings = paramMappings
this._params = params
}
public extractWeights(weights: Float32Array) {
const {
paramMappings,
params
} = extractParams(weights)
this._paramMappings = paramMappings
this._params = params
} }
public forwardInput(input: NetInput): tf.Tensor2D { public forwardInput(input: NetInput): tf.Tensor2D {
const params = this._params
const { params } = this
if (!params) { if (!params) {
throw new Error('FaceLandmarkNet - load model before inference') throw new Error('FaceLandmarkNet - load model before inference')
...@@ -62,20 +39,20 @@ export class FaceLandmarkNet extends NeuralNetwork<NetParams> { ...@@ -62,20 +39,20 @@ export class FaceLandmarkNet extends NeuralNetwork<NetParams> {
return tf.tidy(() => { return tf.tidy(() => {
const batchTensor = input.toBatchTensor(128, true) const batchTensor = input.toBatchTensor(128, true)
let out = conv(batchTensor, params.conv0_params) let out = conv(batchTensor, params.conv0)
out = maxPool(out) out = maxPool(out)
out = conv(out, params.conv1_params) out = conv(out, params.conv1)
out = conv(out, params.conv2_params) out = conv(out, params.conv2)
out = maxPool(out) out = maxPool(out)
out = conv(out, params.conv3_params) out = conv(out, params.conv3)
out = conv(out, params.conv4_params) out = conv(out, params.conv4)
out = maxPool(out) out = maxPool(out)
out = conv(out, params.conv5_params) out = conv(out, params.conv5)
out = conv(out, params.conv6_params) out = conv(out, params.conv6)
out = maxPool(out, [1, 1]) out = maxPool(out, [1, 1])
out = conv(out, params.conv7_params) out = conv(out, params.conv7)
const fc0 = tf.relu(fullyConnectedLayer(out.as2D(out.shape[0], -1), params.fc0_params)) const fc0 = tf.relu(fullyConnectedLayer(out.as2D(out.shape[0], -1), params.fc0))
const fc1 = fullyConnectedLayer(fc0, params.fc1_params) const fc1 = fullyConnectedLayer(fc0, params.fc1)
const createInterleavedTensor = (fillX: number, fillY: number) => const createInterleavedTensor = (fillX: number, fillY: number) =>
tf.stack([ tf.stack([
...@@ -145,4 +122,12 @@ export class FaceLandmarkNet extends NeuralNetwork<NetParams> { ...@@ -145,4 +122,12 @@ export class FaceLandmarkNet extends NeuralNetwork<NetParams> {
? landmarksForBatch ? landmarksForBatch
: landmarksForBatch[0] : landmarksForBatch[0]
} }
protected loadQuantizedParams(uri: string | undefined) {
return loadQuantizedParams(uri)
}
protected extractParams(weights: Float32Array) {
return extractParams(weights)
}
} }
\ No newline at end of file
src/faceLandmarkNet/extractParams.ts
import * as tf from '@tensorflow/tfjs-core'; import * as tf from '@tensorflow/tfjs-core';
import { extractConvParamsFactory } from '../commons/extractConvParamsFactory';
import { extractWeightsFactory } from '../commons/extractWeightsFactory'; import { extractWeightsFactory } from '../commons/extractWeightsFactory';
import { ParamMapping } from '../commons/types'; import { ConvParams, ParamMapping } from '../commons/types';
import { FCParams, NetParams } from './types'; import { FCParams, NetParams } from './types';
export function extractParams(weights: Float32Array): { params: NetParams, paramMappings: ParamMapping[] } { export function extractParams(weights: Float32Array): { params: NetParams, paramMappings: ParamMapping[] } {
const paramMappings: ParamMapping[] = [] const paramMappings: ParamMapping[] = []
const { const {
...@@ -13,9 +13,29 @@ export function extractParams(weights: Float32Array): { params: NetParams, param ...@@ -13,9 +13,29 @@ export function extractParams(weights: Float32Array): { params: NetParams, param
getRemainingWeights getRemainingWeights
} = extractWeightsFactory(weights) } = extractWeightsFactory(weights)
const extractConvParams = extractConvParamsFactory(extractWeights, paramMappings) function extractConvParams(
channelsIn: number,
channelsOut: number,
filterSize: number,
mappedPrefix: string
): ConvParams {
const filters = tf.tensor4d(
extractWeights(channelsIn * channelsOut * filterSize * filterSize),
[filterSize, filterSize, channelsIn, channelsOut]
)
const bias = tf.tensor1d(extractWeights(channelsOut))
paramMappings.push(
{ paramPath: `${mappedPrefix}/filters` },
{ paramPath: `${mappedPrefix}/bias` }
)
return { filters, bias }
}
function extractFcParams(channelsIn: number, channelsOut: number, mappedPrefix: string): FCParams { function extractFcParams(channelsIn: number, channelsOut: number, mappedPrefix: string): FCParams {
const fc_weights = tf.tensor2d(extractWeights(channelsIn * channelsOut), [channelsIn, channelsOut]) const fc_weights = tf.tensor2d(extractWeights(channelsIn * channelsOut), [channelsIn, channelsOut])
const fc_bias = tf.tensor1d(extractWeights(channelsOut)) const fc_bias = tf.tensor1d(extractWeights(channelsOut))
...@@ -30,16 +50,16 @@ export function extractParams(weights: Float32Array): { params: NetParams, param ...@@ -30,16 +50,16 @@ export function extractParams(weights: Float32Array): { params: NetParams, param
} }
} }
const conv0_params = extractConvParams(3, 32, 3, 'conv0_params') const conv0 = extractConvParams(3, 32, 3, 'conv0')
const conv1_params = extractConvParams(32, 64, 3, 'conv1_params') const conv1 = extractConvParams(32, 64, 3, 'conv1')
const conv2_params = extractConvParams(64, 64, 3, 'conv2_params') const conv2 = extractConvParams(64, 64, 3, 'conv2')
const conv3_params = extractConvParams(64, 64, 3, 'conv3_params') const conv3 = extractConvParams(64, 64, 3, 'conv3')
const conv4_params = extractConvParams(64, 64, 3, 'conv4_params') const conv4 = extractConvParams(64, 64, 3, 'conv4')
const conv5_params = extractConvParams(64, 128, 3, 'conv5_params') const conv5 = extractConvParams(64, 128, 3, 'conv5')
const conv6_params = extractConvParams(128, 128, 3, 'conv6_params') const conv6 = extractConvParams(128, 128, 3, 'conv6')
const conv7_params = extractConvParams(128, 256, 3, 'conv7_params') const conv7 = extractConvParams(128, 256, 3, 'conv7')
const fc0_params = extractFcParams(6400, 1024, 'fc0_params') const fc0 = extractFcParams(6400, 1024, 'fc0')
const fc1_params = extractFcParams(1024, 136, 'fc1_params') const fc1 = extractFcParams(1024, 136, 'fc1')
if (getRemainingWeights().length !== 0) { if (getRemainingWeights().length !== 0) {
throw new Error(`weights remaing after extract: ${getRemainingWeights().length}`) throw new Error(`weights remaing after extract: ${getRemainingWeights().length}`)
...@@ -48,16 +68,16 @@ export function extractParams(weights: Float32Array): { params: NetParams, param ...@@ -48,16 +68,16 @@ export function extractParams(weights: Float32Array): { params: NetParams, param
return { return {
paramMappings, paramMappings,
params: { params: {
conv0_params, conv0,
conv1_params, conv1,
conv2_params, conv2,
conv3_params, conv3,
conv4_params, conv4,
conv5_params, conv5,
conv6_params, conv6,
conv7_params, conv7,
fc0_params, fc0,
fc1_params fc1
} }
} }
} }
\ No newline at end of file
src/faceLandmarkNet/loadQuantizedParams.ts
import * as tf from '@tensorflow/tfjs-core'; import * as tf from '@tensorflow/tfjs-core';
import { extractWeightEntry } from '../commons/extractWeightEntry'; import { disposeUnusedWeightTensors } from '../commons/disposeUnusedWeightTensors';
import { extractWeightEntryFactory } from '../commons/extractWeightEntryFactory';
import { loadWeightMap } from '../commons/loadWeightMap'; import { loadWeightMap } from '../commons/loadWeightMap';
import { ConvParams, ParamMapping } from '../commons/types'; import { ConvParams, ParamMapping } from '../commons/types';
import { FCParams, NetParams } from './types'; import { FCParams, NetParams } from './types';
...@@ -9,30 +10,20 @@ const DEFAULT_MODEL_NAME = 'face_landmark_68_model' ...@@ -9,30 +10,20 @@ const DEFAULT_MODEL_NAME = 'face_landmark_68_model'
function extractorsFactory(weightMap: any, paramMappings: ParamMapping[]) { function extractorsFactory(weightMap: any, paramMappings: ParamMapping[]) {
const extractWeightEntry = extractWeightEntryFactory(weightMap, paramMappings)
function extractConvParams(prefix: string, mappedPrefix: string): ConvParams { function extractConvParams(prefix: string, mappedPrefix: string): ConvParams {
const filtersEntry = extractWeightEntry(weightMap, `${prefix}/kernel`, 4) const filters = extractWeightEntry<tf.Tensor4D>(`${prefix}/kernel`, 4, `${mappedPrefix}/filters`)
const biasEntry = extractWeightEntry(weightMap, `${prefix}/bias`, 1) const bias = extractWeightEntry<tf.Tensor1D>(`${prefix}/bias`, 1, `${mappedPrefix}/bias`)
paramMappings.push(
{ originalPath: filtersEntry.path, paramPath: `${mappedPrefix}/filters` }, return { filters, bias }
{ originalPath: biasEntry.path, paramPath: `${mappedPrefix}/bias` }
)
return {
filters: filtersEntry.tensor as tf.Tensor4D,
bias: biasEntry.tensor as tf.Tensor1D
}
} }
function extractFcParams(prefix: string, mappedPrefix: string): FCParams { function extractFcParams(prefix: string, mappedPrefix: string): FCParams {
const weightsEntry = extractWeightEntry(weightMap, `${prefix}/kernel`, 2) const weights = extractWeightEntry<tf.Tensor2D>(`${prefix}/kernel`, 2, `${mappedPrefix}/weights`)
const biasEntry = extractWeightEntry(weightMap, `${prefix}/bias`, 1) const bias = extractWeightEntry<tf.Tensor1D>(`${prefix}/bias`, 1, `${mappedPrefix}/bias`)
paramMappings.push(
{ originalPath: weightsEntry.path, paramPath: `${mappedPrefix}/weights` }, return { weights, bias }
{ originalPath: biasEntry.path, paramPath: `${mappedPrefix}/bias` }
)
return {
weights: weightsEntry.tensor as tf.Tensor2D,
bias: biasEntry.tensor as tf.Tensor1D
}
} }
return { return {
...@@ -54,17 +45,19 @@ export async function loadQuantizedParams( ...@@ -54,17 +45,19 @@ export async function loadQuantizedParams(
} = extractorsFactory(weightMap, paramMappings) } = extractorsFactory(weightMap, paramMappings)
const params = { const params = {
conv0_params: extractConvParams('conv2d_0', 'conv0_params'), conv0: extractConvParams('conv2d_0', 'conv0'),
conv1_params: extractConvParams('conv2d_1', 'conv1_params'), conv1: extractConvParams('conv2d_1', 'conv1'),
conv2_params: extractConvParams('conv2d_2', 'conv2_params'), conv2: extractConvParams('conv2d_2', 'conv2'),
conv3_params: extractConvParams('conv2d_3', 'conv3_params'), conv3: extractConvParams('conv2d_3', 'conv3'),
conv4_params: extractConvParams('conv2d_4', 'conv4_params'), conv4: extractConvParams('conv2d_4', 'conv4'),
conv5_params: extractConvParams('conv2d_5', 'conv5_params'), conv5: extractConvParams('conv2d_5', 'conv5'),
conv6_params: extractConvParams('conv2d_6', 'conv6_params'), conv6: extractConvParams('conv2d_6', 'conv6'),
conv7_params: extractConvParams('conv2d_7', 'conv7_params'), conv7: extractConvParams('conv2d_7', 'conv7'),
fc0_params: extractFcParams('dense', 'fc0_params'), fc0: extractFcParams('dense', 'fc0'),
fc1_params: extractFcParams('logits', 'fc1_params') fc1: extractFcParams('logits', 'fc1')
} }
disposeUnusedWeightTensors(weightMap, paramMappings)
return { params, paramMappings } return { params, paramMappings }
} }
\ No newline at end of file
src/faceLandmarkNet/types.ts
...@@ -8,14 +8,14 @@ export type FCParams = { ...@@ -8,14 +8,14 @@ export type FCParams = {
} }
export type NetParams = { export type NetParams = {
conv0_params: ConvParams conv0: ConvParams
conv1_params: ConvParams conv1: ConvParams
conv2_params: ConvParams conv2: ConvParams
conv3_params: ConvParams conv3: ConvParams
conv4_params: ConvParams conv4: ConvParams
conv5_params: ConvParams conv5: ConvParams
conv6_params: ConvParams conv6: ConvParams
conv7_params: ConvParams conv7: ConvParams
fc0_params: FCParams fc0: FCParams
fc1_params: FCParams fc1: FCParams
} }
\ No newline at end of file
src/faceRecognitionNet/FaceRecognitionNet.ts
import * as tf from '@tensorflow/tfjs-core'; import * as tf from '@tensorflow/tfjs-core';
import { NeuralNetwork } from '../commons/NeuralNetwork';
import { NetInput } from '../NetInput'; import { NetInput } from '../NetInput';
import { toNetInput } from '../toNetInput'; import { toNetInput } from '../toNetInput';
import { TNetInput } from '../types'; import { TNetInput } from '../types';
...@@ -10,28 +11,17 @@ import { normalize } from './normalize'; ...@@ -10,28 +11,17 @@ import { normalize } from './normalize';
import { residual, residualDown } from './residualLayer'; import { residual, residualDown } from './residualLayer';
import { NetParams } from './types'; import { NetParams } from './types';
export class FaceRecognitionNet { export class FaceRecognitionNet extends NeuralNetwork<NetParams> {
private _params: NetParams constructor() {
super('FaceRecognitionNet')
public async load(weightsOrUrl: Float32Array | string | undefined): Promise<void> {
if (weightsOrUrl instanceof Float32Array) {
this.extractWeights(weightsOrUrl)
return
} }
if (weightsOrUrl && typeof weightsOrUrl !== 'string') { public forwardInput(input: NetInput): tf.Tensor2D {
throw new Error('FaceLandmarkNet.load - expected model uri, or weights as Float32Array')
}
this._params = await loadQuantizedParams(weightsOrUrl)
}
public extractWeights(weights: Float32Array) { const { params } = this
this._params = extractParams(weights)
}
public forwardInput(input: NetInput): tf.Tensor2D { if (!params) {
if (!this._params) {
throw new Error('FaceRecognitionNet - load model before inference') throw new Error('FaceRecognitionNet - load model before inference')
} }
...@@ -40,29 +30,29 @@ export class FaceRecognitionNet { ...@@ -40,29 +30,29 @@ export class FaceRecognitionNet {
const normalized = normalize(batchTensor) const normalized = normalize(batchTensor)
let out = convDown(normalized, this._params.conv32_down) let out = convDown(normalized, params.conv32_down)
out = tf.maxPool(out, 3, 2, 'valid') out = tf.maxPool(out, 3, 2, 'valid')
out = residual(out, this._params.conv32_1) out = residual(out, params.conv32_1)
out = residual(out, this._params.conv32_2) out = residual(out, params.conv32_2)
out = residual(out, this._params.conv32_3) out = residual(out, params.conv32_3)
out = residualDown(out, this._params.conv64_down) out = residualDown(out, params.conv64_down)
out = residual(out, this._params.conv64_1) out = residual(out, params.conv64_1)
out = residual(out, this._params.conv64_2) out = residual(out, params.conv64_2)
out = residual(out, this._params.conv64_3) out = residual(out, params.conv64_3)
out = residualDown(out, this._params.conv128_down) out = residualDown(out, params.conv128_down)
out = residual(out, this._params.conv128_1) out = residual(out, params.conv128_1)
out = residual(out, this._params.conv128_2) out = residual(out, params.conv128_2)
out = residualDown(out, this._params.conv256_down) out = residualDown(out, params.conv256_down)
out = residual(out, this._params.conv256_1) out = residual(out, params.conv256_1)
out = residual(out, this._params.conv256_2) out = residual(out, params.conv256_2)
out = residualDown(out, this._params.conv256_down_out) out = residualDown(out, params.conv256_down_out)
const globalAvg = out.mean([1, 2]) as tf.Tensor2D const globalAvg = out.mean([1, 2]) as tf.Tensor2D
const fullyConnected = tf.matMul(globalAvg, this._params.fc) const fullyConnected = tf.matMul(globalAvg, params.fc)
return fullyConnected return fullyConnected
}) })
...@@ -89,4 +79,12 @@ export class FaceRecognitionNet { ...@@ -89,4 +79,12 @@ export class FaceRecognitionNet {
? faceDescriptorsForBatch ? faceDescriptorsForBatch
: faceDescriptorsForBatch[0] : faceDescriptorsForBatch[0]
} }
protected loadQuantizedParams(uri: string | undefined) {
return loadQuantizedParams(uri)
}
protected extractParams(weights: Float32Array) {
return extractParams(weights)
}
} }
\ No newline at end of file
src/faceRecognitionNet/extractParams.ts
import * as tf from '@tensorflow/tfjs-core'; import * as tf from '@tensorflow/tfjs-core';
import { extractWeightsFactory } from '../commons/extractWeightsFactory'; import { extractWeightsFactory } from '../commons/extractWeightsFactory';
import { ExtractWeightsFunction } from '../commons/types'; import { ConvParams, ExtractWeightsFunction, ParamMapping } from '../commons/types';
import { isFloat } from '../utils'; import { isFloat } from '../utils';
import { ConvLayerParams, NetParams, ResidualLayerParams, ScaleLayerParams } from './types'; import { ConvLayerParams, NetParams, ResidualLayerParams, ScaleLayerParams } from './types';
function extractorsFactory(extractWeights: ExtractWeightsFunction) { function extractorsFactory(extractWeights: ExtractWeightsFunction, paramMappings: ParamMapping[]) {
function extractFilterValues(numFilterValues: number, numFilters: number, filterSize: number): tf.Tensor4D { function extractFilterValues(numFilterValues: number, numFilters: number, filterSize: number): tf.Tensor4D {
const weights = extractWeights(numFilterValues) const weights = extractWeights(numFilterValues)
...@@ -15,15 +15,42 @@ function extractorsFactory(extractWeights: ExtractWeightsFunction) { ...@@ -15,15 +15,42 @@ function extractorsFactory(extractWeights: ExtractWeightsFunction) {
throw new Error(`depth has to be an integer: ${depth}, weights.length: ${weights.length}, numFilters: ${numFilters}, filterSize: ${filterSize}`) throw new Error(`depth has to be an integer: ${depth}, weights.length: ${weights.length}, numFilters: ${numFilters}, filterSize: ${filterSize}`)
} }
return tf.transpose( return tf.tidy(
() => tf.transpose(
tf.tensor4d(weights, [numFilters, depth, filterSize, filterSize]), tf.tensor4d(weights, [numFilters, depth, filterSize, filterSize]),
[2, 3, 1, 0] [2, 3, 1, 0]
) )
)
} }
function extractScaleLayerParams(numWeights: number): ScaleLayerParams { function extractConvParams(
numFilterValues: number,
numFilters: number,
filterSize: number,
mappedPrefix: string
): ConvParams {
const filters = extractFilterValues(numFilterValues, numFilters, filterSize)
const bias = tf.tensor1d(extractWeights(numFilters))
paramMappings.push(
{ paramPath: `${mappedPrefix}/filters` },
{ paramPath: `${mappedPrefix}/bias` }
)
return { filters, bias }
}
function extractScaleLayerParams(numWeights: number, mappedPrefix: string): ScaleLayerParams {
const weights = tf.tensor1d(extractWeights(numWeights)) const weights = tf.tensor1d(extractWeights(numWeights))
const biases = tf.tensor1d(extractWeights(numWeights)) const biases = tf.tensor1d(extractWeights(numWeights))
paramMappings.push(
{ paramPath: `${mappedPrefix}/weights` },
{ paramPath: `${mappedPrefix}/biases` }
)
return { return {
weights, weights,
biases biases
...@@ -33,34 +60,28 @@ function extractorsFactory(extractWeights: ExtractWeightsFunction) { ...@@ -33,34 +60,28 @@ function extractorsFactory(extractWeights: ExtractWeightsFunction) {
function extractConvLayerParams( function extractConvLayerParams(
numFilterValues: number, numFilterValues: number,
numFilters: number, numFilters: number,
filterSize: number filterSize: number,
mappedPrefix: string
): ConvLayerParams { ): ConvLayerParams {
const conv_filters = extractFilterValues(numFilterValues, numFilters, filterSize)
const conv_bias = tf.tensor1d(extractWeights(numFilters))
const scale = extractScaleLayerParams(numFilters)
return { const conv = extractConvParams(numFilterValues, numFilters, filterSize, `${mappedPrefix}/conv`)
conv: { const scale = extractScaleLayerParams(numFilters, `${mappedPrefix}/scale`)
filters: conv_filters,
bias: conv_bias return { conv, scale }
},
scale
}
} }
function extractResidualLayerParams( function extractResidualLayerParams(
numFilterValues: number, numFilterValues: number,
numFilters: number, numFilters: number,
filterSize: number, filterSize: number,
mappedPrefix: string,
isDown: boolean = false isDown: boolean = false
): ResidualLayerParams { ): ResidualLayerParams {
const conv1: ConvLayerParams = extractConvLayerParams((isDown ? 0.5 : 1) * numFilterValues, numFilters, filterSize)
const conv2: ConvLayerParams = extractConvLayerParams(numFilterValues, numFilters, filterSize)
return { const conv1 = extractConvLayerParams((isDown ? 0.5 : 1) * numFilterValues, numFilters, filterSize, `${mappedPrefix}/conv1`)
conv1, const conv2 = extractConvLayerParams(numFilterValues, numFilters, filterSize, `${mappedPrefix}/conv2`)
conv2
} return { conv1, conv2 }
} }
return { return {
...@@ -70,43 +91,49 @@ function extractorsFactory(extractWeights: ExtractWeightsFunction) { ...@@ -70,43 +91,49 @@ function extractorsFactory(extractWeights: ExtractWeightsFunction) {
} }
export function extractParams(weights: Float32Array): NetParams { export function extractParams(weights: Float32Array): { params: NetParams, paramMappings: ParamMapping[] } {
const { const {
extractWeights, extractWeights,
getRemainingWeights getRemainingWeights
} = extractWeightsFactory(weights) } = extractWeightsFactory(weights)
const paramMappings: ParamMapping[] = []
const { const {
extractConvLayerParams, extractConvLayerParams,
extractResidualLayerParams extractResidualLayerParams
} = extractorsFactory(extractWeights) } = extractorsFactory(extractWeights, paramMappings)
const conv32_down = extractConvLayerParams(4704, 32, 7) const conv32_down = extractConvLayerParams(4704, 32, 7, 'conv32_down')
const conv32_1 = extractResidualLayerParams(9216, 32, 3) const conv32_1 = extractResidualLayerParams(9216, 32, 3, 'conv32_1')
const conv32_2 = extractResidualLayerParams(9216, 32, 3) const conv32_2 = extractResidualLayerParams(9216, 32, 3, 'conv32_2')
const conv32_3 = extractResidualLayerParams(9216, 32, 3) const conv32_3 = extractResidualLayerParams(9216, 32, 3, 'conv32_3')
const conv64_down = extractResidualLayerParams(36864, 64, 3, true) const conv64_down = extractResidualLayerParams(36864, 64, 3, 'conv64_down', true)
const conv64_1 = extractResidualLayerParams(36864, 64, 3) const conv64_1 = extractResidualLayerParams(36864, 64, 3, 'conv64_1')
const conv64_2 = extractResidualLayerParams(36864, 64, 3) const conv64_2 = extractResidualLayerParams(36864, 64, 3, 'conv64_2')
const conv64_3 = extractResidualLayerParams(36864, 64, 3) const conv64_3 = extractResidualLayerParams(36864, 64, 3, 'conv64_3')
const conv128_down = extractResidualLayerParams(147456, 128, 3, true) const conv128_down = extractResidualLayerParams(147456, 128, 3, 'conv128_down', true)
const conv128_1 = extractResidualLayerParams(147456, 128, 3) const conv128_1 = extractResidualLayerParams(147456, 128, 3, 'conv128_1')
const conv128_2 = extractResidualLayerParams(147456, 128, 3) const conv128_2 = extractResidualLayerParams(147456, 128, 3, 'conv128_2')
const conv256_down = extractResidualLayerParams(589824, 256, 3, true) const conv256_down = extractResidualLayerParams(589824, 256, 3, 'conv256_down', true)
const conv256_1 = extractResidualLayerParams(589824, 256, 3) const conv256_1 = extractResidualLayerParams(589824, 256, 3, 'conv256_1')
const conv256_2 = extractResidualLayerParams(589824, 256, 3) const conv256_2 = extractResidualLayerParams(589824, 256, 3, 'conv256_2')
const conv256_down_out = extractResidualLayerParams(589824, 256, 3) const conv256_down_out = extractResidualLayerParams(589824, 256, 3, 'conv256_down_out')
const fc = tf.transpose(tf.tensor2d(extractWeights(256 * 128), [128, 256]), [1, 0]) const fc = tf.tidy(
() => tf.transpose(tf.tensor2d(extractWeights(256 * 128), [128, 256]), [1, 0])
)
paramMappings.push({ paramPath: `fc` })
if (getRemainingWeights().length !== 0) { if (getRemainingWeights().length !== 0) {
throw new Error(`weights remaing after extract: ${getRemainingWeights().length}`) throw new Error(`weights remaing after extract: ${getRemainingWeights().length}`)
} }
return { const params = {
conv32_down, conv32_down,
conv32_1, conv32_1,
conv32_2, conv32_2,
...@@ -124,4 +151,6 @@ export function extractParams(weights: Float32Array): NetParams { ...@@ -124,4 +151,6 @@ export function extractParams(weights: Float32Array): NetParams {
conv256_down_out, conv256_down_out,
fc fc
} }
return { params, paramMappings }
} }
\ No newline at end of file
src/faceRecognitionNet/loadQuantizedParams.ts
import { isTensor1D, isTensor2D, isTensor4D } from '../commons/isTensor'; import * as tf from '@tensorflow/tfjs-core';
import { disposeUnusedWeightTensors } from '../commons/disposeUnusedWeightTensors';
import { extractWeightEntryFactory } from '../commons/extractWeightEntryFactory';
import { isTensor2D } from '../commons/isTensor';
import { loadWeightMap } from '../commons/loadWeightMap'; import { loadWeightMap } from '../commons/loadWeightMap';
import { ConvLayerParams, ResidualLayerParams, ScaleLayerParams } from './types'; import { ParamMapping } from '../commons/types';
import { ConvLayerParams, NetParams, ResidualLayerParams, ScaleLayerParams } from './types';
const DEFAULT_MODEL_NAME = 'face_recognition_model' const DEFAULT_MODEL_NAME = 'face_recognition_model'
function extractorsFactory(weightMap: any) { function extractorsFactory(weightMap: any, paramMappings: ParamMapping[]) {
function extractScaleLayerParams(prefix: string): ScaleLayerParams { const extractWeightEntry = extractWeightEntryFactory(weightMap, paramMappings)
const params = {
weights: weightMap[`${prefix}/scale/weights`],
biases: weightMap[`${prefix}/scale/biases`]
}
if (!isTensor1D(params.weights)) { function extractScaleLayerParams(prefix: string): ScaleLayerParams {
throw new Error(`expected weightMap[${prefix}/scale/weights] to be a Tensor1D, instead have ${params.weights}`)
}
if (!isTensor1D(params.biases)) { const weights = extractWeightEntry<tf.Tensor1D>(`${prefix}/scale/weights`, 1)
throw new Error(`expected weightMap[${prefix}/scale/biases] to be a Tensor1D, instead have ${params.biases}`) const biases = extractWeightEntry<tf.Tensor1D>(`${prefix}/scale/biases`, 1)
}
return params return { weights, biases }
} }
function extractConvLayerParams(prefix: string): ConvLayerParams { function extractConvLayerParams(prefix: string): ConvLayerParams {
const params = {
filters: weightMap[`${prefix}/conv/filters`],
bias: weightMap[`${prefix}/conv/bias`]
}
if (!isTensor4D(params.filters)) { const filters = extractWeightEntry<tf.Tensor4D>(`${prefix}/conv/filters`, 4)
throw new Error(`expected weightMap[${prefix}/conv/filters] to be a Tensor1D, instead have ${params.filters}`) const bias = extractWeightEntry<tf.Tensor1D>(`${prefix}/conv/bias`, 1)
} const scale = extractScaleLayerParams(prefix)
if (!isTensor1D(params.bias)) { return { conv: { filters, bias }, scale }
throw new Error(`expected weightMap[${prefix}/conv/bias] to be a Tensor1D, instead have ${params.bias}`)
}
return {
conv: params,
scale: extractScaleLayerParams(prefix)
}
} }
function extractResidualLayerParams(prefix: string): ResidualLayerParams { function extractResidualLayerParams(prefix: string): ResidualLayerParams {
...@@ -57,13 +44,17 @@ function extractorsFactory(weightMap: any) { ...@@ -57,13 +44,17 @@ function extractorsFactory(weightMap: any) {
} }
export async function loadQuantizedParams(uri: string | undefined): Promise<any> { export async function loadQuantizedParams(
uri: string | undefined
): Promise<{ params: NetParams, paramMappings: ParamMapping[] }> {
const weightMap = await loadWeightMap(uri, DEFAULT_MODEL_NAME) const weightMap = await loadWeightMap(uri, DEFAULT_MODEL_NAME)
const paramMappings: ParamMapping[] = []
const { const {
extractConvLayerParams, extractConvLayerParams,
extractResidualLayerParams extractResidualLayerParams
} = extractorsFactory(weightMap) } = extractorsFactory(weightMap, paramMappings)
const conv32_down = extractConvLayerParams('conv32_down') const conv32_down = extractConvLayerParams('conv32_down')
const conv32_1 = extractResidualLayerParams('conv32_1') const conv32_1 = extractResidualLayerParams('conv32_1')
...@@ -85,12 +76,13 @@ export async function loadQuantizedParams(uri: string | undefined): Promise<any> ...@@ -85,12 +76,13 @@ export async function loadQuantizedParams(uri: string | undefined): Promise<any>
const conv256_down_out = extractResidualLayerParams('conv256_down_out') const conv256_down_out = extractResidualLayerParams('conv256_down_out')
const fc = weightMap['fc'] const fc = weightMap['fc']
paramMappings.push({ originalPath: 'fc', paramPath: 'fc' })
if (!isTensor2D(fc)) { if (!isTensor2D(fc)) {
throw new Error(`expected weightMap[fc] to be a Tensor2D, instead have ${fc}`) throw new Error(`expected weightMap[fc] to be a Tensor2D, instead have ${fc}`)
} }
return { const params = {
conv32_down, conv32_down,
conv32_1, conv32_1,
conv32_2, conv32_2,
...@@ -108,4 +100,8 @@ export async function loadQuantizedParams(uri: string | undefined): Promise<any> ...@@ -108,4 +100,8 @@ export async function loadQuantizedParams(uri: string | undefined): Promise<any>
conv256_down_out, conv256_down_out,
fc fc
} }
disposeUnusedWeightTensors(weightMap, paramMappings)
return { params, paramMappings }
} }
\ No newline at end of file
test/tests/commons/NeuralNetwork.test.ts
import { NeuralNetwork } from '../../../src/commons/NeuralNetwork';
import * as tf from '@tensorflow/tfjs-core'; import * as tf from '@tensorflow/tfjs-core';
import { NeuralNetwork } from '../../../src/commons/NeuralNetwork';
class FakeNeuralNetwork extends NeuralNetwork<any> { class FakeNeuralNetwork extends NeuralNetwork<any> {
constructor( constructor(
convFilter: tf.Tensor = tf.tensor(0), convFilter: tf.Tensor = tf.tensor(0),
convBias: tf.Tensor = tf.tensor(0), convBias: tf.Tensor = tf.tensor(0),
fcWeights: tf.Tensor = tf.tensor(0) fcWeights: tf.Tensor = tf.tensor(0)
) { ) {
super() super('FakeNeuralNetwork')
this._params = { this._params = {
conv: { conv: {
filter: convFilter, filter: convFilter,
......
test/tests/e2e/faceDetectionNet.test.ts
import * as faceapi from '../../../src'; import * as faceapi from '../../../src';
import { FaceDetection } from '../../../src/faceDetectionNet/FaceDetection'; import { FaceDetection } from '../../../src/faceDetectionNet/FaceDetection';
import { IRect } from '../../../src/Rect'; import { IRect } from '../../../src/Rect';
import { expectMaxDelta } from '../../utils'; import { describeWithNets, expectAllTensorsReleased, expectMaxDelta } from '../../utils';
function expectRectClose( function expectRectClose(
result: IRect, result: IRect,
...@@ -33,19 +33,11 @@ describe('faceDetectionNet', () => { ...@@ -33,19 +33,11 @@ describe('faceDetectionNet', () => {
imgEl = await faceapi.bufferToImage(img) imgEl = await faceapi.bufferToImage(img)
}) })
describe('uncompressed weights', () => { describeWithNets('uncompressed weights', { withFaceDetectionNet: { quantized: false } }, ({ faceDetectionNet }) => {
let faceDetectionNet: faceapi.FaceDetectionNet
const expectedScores = [0.98, 0.89, 0.82, 0.75, 0.58, 0.55] const expectedScores = [0.98, 0.89, 0.82, 0.75, 0.58, 0.55]
const maxBoxDelta = 1 const maxBoxDelta = 1
beforeAll(async () => {
const res = await fetch('base/weights/uncompressed/face_detection_model.weights')
const weights = new Float32Array(await res.arrayBuffer())
faceDetectionNet = faceapi.faceDetectionNet(weights)
})
it('scores > 0.8', async () => { it('scores > 0.8', async () => {
const detections = await faceDetectionNet.locateFaces(imgEl) as FaceDetection[] const detections = await faceDetectionNet.locateFaces(imgEl) as FaceDetection[]
...@@ -72,18 +64,11 @@ describe('faceDetectionNet', () => { ...@@ -72,18 +64,11 @@ describe('faceDetectionNet', () => {
}) })
describe('quantized weights', () => { describeWithNets('quantized weights', { withFaceDetectionNet: { quantized: true } }, ({ faceDetectionNet }) => {
let faceDetectionNet: faceapi.FaceDetectionNet
const expectedScores = [0.97, 0.88, 0.83, 0.82, 0.59, 0.52] const expectedScores = [0.97, 0.88, 0.83, 0.82, 0.59, 0.52]
const maxBoxDelta = 5 const maxBoxDelta = 5
beforeAll(async () => {
faceDetectionNet = new faceapi.FaceDetectionNet()
await faceDetectionNet.load('base/weights')
})
it('scores > 0.8', async () => { it('scores > 0.8', async () => {
const detections = await faceDetectionNet.locateFaces(imgEl) as FaceDetection[] const detections = await faceDetectionNet.locateFaces(imgEl) as FaceDetection[]
...@@ -110,4 +95,33 @@ describe('faceDetectionNet', () => { ...@@ -110,4 +95,33 @@ describe('faceDetectionNet', () => {
}) })
describe('no memory leaks', () => {
describe('NeuralNetwork, uncompressed model', () => {
it('disposes all param tensors', async () => {
await expectAllTensorsReleased(async () => {
const res = await fetch('base/weights/uncompressed/face_detection_model.weights')
const weights = new Float32Array(await res.arrayBuffer())
const net = faceapi.faceDetectionNet(weights)
net.dispose()
})
})
})
describe('NeuralNetwork, quantized model', () => {
it('disposes all param tensors', async () => {
await expectAllTensorsReleased(async () => {
const net = new faceapi.FaceDetectionNet()
await net.load('base/weights')
net.dispose()
})
})
})
})
}) })
\ No newline at end of file
test/tests/e2e/faceLandmarkNet.test.ts
...@@ -5,7 +5,7 @@ import { isTensor3D } from '../../../src/commons/isTensor'; ...@@ -5,7 +5,7 @@ import { isTensor3D } from '../../../src/commons/isTensor';
import { FaceLandmarks } from '../../../src/faceLandmarkNet/FaceLandmarks'; import { FaceLandmarks } from '../../../src/faceLandmarkNet/FaceLandmarks';
import { Point } from '../../../src/Point'; import { Point } from '../../../src/Point';
import { Dimensions, TMediaElement } from '../../../src/types'; import { Dimensions, TMediaElement } from '../../../src/types';
import { expectMaxDelta, expectAllTensorsReleased, tensor3D } from '../../utils'; import { expectMaxDelta, expectAllTensorsReleased, tensor3D, describeWithNets } from '../../utils';
import { NetInput } from '../../../src/NetInput'; import { NetInput } from '../../../src/NetInput';
import { toNetInput } from '../../../src'; import { toNetInput } from '../../../src';
...@@ -38,15 +38,7 @@ describe('faceLandmarkNet', () => { ...@@ -38,15 +38,7 @@ describe('faceLandmarkNet', () => {
faceLandmarkPositionsRect = await (await fetch('base/test/data/faceLandmarkPositionsRect.json')).json() faceLandmarkPositionsRect = await (await fetch('base/test/data/faceLandmarkPositionsRect.json')).json()
}) })
describe('uncompressed weights', () => { describeWithNets('uncompressed weights', { withFaceLandmarkNet: { quantized: false } }, ({ faceLandmarkNet }) => {
let faceLandmarkNet: faceapi.FaceLandmarkNet
beforeAll(async () => {
const res = await fetch('base/weights/uncompressed/face_landmark_68_model.weights')
const weights = new Float32Array(await res.arrayBuffer())
faceLandmarkNet = faceapi.faceLandmarkNet(weights)
})
it('computes face landmarks for squared input', async () => { it('computes face landmarks for squared input', async () => {
const { width, height } = imgEl1 const { width, height } = imgEl1
...@@ -78,14 +70,7 @@ describe('faceLandmarkNet', () => { ...@@ -78,14 +70,7 @@ describe('faceLandmarkNet', () => {
}) })
describe('quantized weights', () => { describeWithNets('quantized weights', { withFaceLandmarkNet: { quantized: true } }, ({ faceLandmarkNet }) => {
let faceLandmarkNet: faceapi.FaceLandmarkNet
beforeAll(async () => {
faceLandmarkNet = new faceapi.FaceLandmarkNet()
await faceLandmarkNet.load('base/weights')
})
it('computes face landmarks for squared input', async () => { it('computes face landmarks for squared input', async () => {
const { width, height } = imgEl1 const { width, height } = imgEl1
...@@ -117,15 +102,7 @@ describe('faceLandmarkNet', () => { ...@@ -117,15 +102,7 @@ describe('faceLandmarkNet', () => {
}) })
describe('batch inputs', () => { describeWithNets('batch inputs', { withFaceLandmarkNet: { quantized: false } }, ({ faceLandmarkNet }) => {
let faceLandmarkNet: faceapi.FaceLandmarkNet
beforeAll(async () => {
const res = await fetch('base/weights/uncompressed/face_landmark_68_model.weights')
const weights = new Float32Array(await res.arrayBuffer())
faceLandmarkNet = faceapi.faceLandmarkNet(weights)
})
it('computes face landmarks for batch of image elements', async () => { it('computes face landmarks for batch of image elements', async () => {
const inputs = [imgEl1, imgEl2, imgElRect] const inputs = [imgEl1, imgEl2, imgElRect]
...@@ -229,13 +206,31 @@ describe('faceLandmarkNet', () => { ...@@ -229,13 +206,31 @@ describe('faceLandmarkNet', () => {
}) })
describe('no memory leaks', () => { describeWithNets('no memory leaks', { withFaceLandmarkNet: { quantized: true } }, ({ faceLandmarkNet }) => {
let faceLandmarkNet: faceapi.FaceLandmarkNet describe('NeuralNetwork, uncompressed model', () => {
it('disposes all param tensors', async () => {
await expectAllTensorsReleased(async () => {
const res = await fetch('base/weights/uncompressed/face_landmark_68_model.weights')
const weights = new Float32Array(await res.arrayBuffer())
const net = faceapi.faceLandmarkNet(weights)
net.dispose()
})
})
})
describe('NeuralNetwork, quantized model', () => {
it('disposes all param tensors', async () => {
await expectAllTensorsReleased(async () => {
const net = new faceapi.FaceLandmarkNet()
await net.load('base/weights')
net.dispose()
})
})
beforeAll(async () => {
faceLandmarkNet = new faceapi.FaceLandmarkNet()
await faceLandmarkNet.load('base/weights')
}) })
describe('forwardInput', () => { describe('forwardInput', () => {
......
test/tests/e2e/faceRecognitionNet.test.ts
...@@ -2,7 +2,7 @@ import * as tf from '@tensorflow/tfjs-core'; ...@@ -2,7 +2,7 @@ import * as tf from '@tensorflow/tfjs-core';
import * as faceapi from '../../../src'; import * as faceapi from '../../../src';
import { NetInput } from '../../../src/NetInput'; import { NetInput } from '../../../src/NetInput';
import { expectAllTensorsReleased } from '../../utils'; import { expectAllTensorsReleased, describeWithNets } from '../../utils';
import { toNetInput } from '../../../src'; import { toNetInput } from '../../../src';
describe('faceRecognitionNet', () => { describe('faceRecognitionNet', () => {
...@@ -26,15 +26,7 @@ describe('faceRecognitionNet', () => { ...@@ -26,15 +26,7 @@ describe('faceRecognitionNet', () => {
faceDescriptorRect = await (await fetch('base/test/data/faceDescriptorRect.json')).json() faceDescriptorRect = await (await fetch('base/test/data/faceDescriptorRect.json')).json()
}) })
describe('uncompressed weights', () => { describeWithNets('uncompressed weights', { withFaceRecognitionNet: { quantized: false } }, ({ faceRecognitionNet }) => {
let faceRecognitionNet: faceapi.FaceRecognitionNet
beforeAll(async () => {
const res = await fetch('base/weights/uncompressed/face_recognition_model.weights')
const weights = new Float32Array(await res.arrayBuffer())
faceRecognitionNet = faceapi.faceRecognitionNet(weights)
})
it('computes face descriptor for squared input', async () => { it('computes face descriptor for squared input', async () => {
const result = await faceRecognitionNet.computeFaceDescriptor(imgEl1) as Float32Array const result = await faceRecognitionNet.computeFaceDescriptor(imgEl1) as Float32Array
...@@ -52,14 +44,7 @@ describe('faceRecognitionNet', () => { ...@@ -52,14 +44,7 @@ describe('faceRecognitionNet', () => {
// TODO: figure out why descriptors return NaN in the test cases // TODO: figure out why descriptors return NaN in the test cases
/* /*
describe('quantized weights', () => { describeWithNets('quantized weights', { withFaceRecognitionNet: { quantized: true } }, ({ faceRecognitionNet }) => {
let faceRecognitionNet: faceapi.FaceRecognitionNet
beforeAll(async () => {
faceRecognitionNet = new faceapi.FaceRecognitionNet()
await faceRecognitionNet.load('base/weights')
})
it('computes face descriptor for squared input', async () => { it('computes face descriptor for squared input', async () => {
const result = await faceRecognitionNet.computeFaceDescriptor(imgEl1) as Float32Array const result = await faceRecognitionNet.computeFaceDescriptor(imgEl1) as Float32Array
...@@ -76,15 +61,7 @@ describe('faceRecognitionNet', () => { ...@@ -76,15 +61,7 @@ describe('faceRecognitionNet', () => {
}) })
*/ */
describe('batch inputs', () => { describeWithNets('batch inputs', { withFaceRecognitionNet: { quantized: false } }, ({ faceRecognitionNet }) => {
let faceRecognitionNet: faceapi.FaceRecognitionNet
beforeAll(async () => {
const res = await fetch('base/weights/uncompressed/face_recognition_model.weights')
const weights = new Float32Array(await res.arrayBuffer())
faceRecognitionNet = faceapi.faceRecognitionNet(weights)
})
it('computes face descriptors for batch of image elements', async () => { it('computes face descriptors for batch of image elements', async () => {
const inputs = [imgEl1, imgEl2, imgElRect] const inputs = [imgEl1, imgEl2, imgElRect]
...@@ -156,14 +133,31 @@ describe('faceRecognitionNet', () => { ...@@ -156,14 +133,31 @@ describe('faceRecognitionNet', () => {
}) })
describe('no memory leaks', () => { describeWithNets('no memory leaks', { withFaceRecognitionNet: { quantized: false } }, ({ faceRecognitionNet }) => {
let faceRecognitionNet: faceapi.FaceRecognitionNet describe('NeuralNetwork, uncompressed model', () => {
beforeAll(async () => { it('disposes all param tensors', async () => {
await expectAllTensorsReleased(async () => {
const res = await fetch('base/weights/uncompressed/face_recognition_model.weights') const res = await fetch('base/weights/uncompressed/face_recognition_model.weights')
const weights = new Float32Array(await res.arrayBuffer()) const weights = new Float32Array(await res.arrayBuffer())
faceRecognitionNet = faceapi.faceRecognitionNet(weights) const net = faceapi.faceRecognitionNet(weights)
net.dispose()
})
})
})
describe('NeuralNetwork, quantized model', () => {
it('disposes all param tensors', async () => {
await expectAllTensorsReleased(async () => {
const net = new faceapi.FaceRecognitionNet()
await net.load('base/weights')
net.dispose()
})
})
}) })
describe('forwardInput', () => { describe('forwardInput', () => {
...@@ -292,5 +286,4 @@ describe('faceRecognitionNet', () => { ...@@ -292,5 +286,4 @@ describe('faceRecognitionNet', () => {
}) })
}) })
}) })
\ No newline at end of file
test/utils.ts
import * as tf from '@tensorflow/tfjs-core'; import * as tf from '@tensorflow/tfjs-core';
import * as faceapi from '../src/';
import { NeuralNetwork } from '../src/commons/NeuralNetwork';
export function zeros(length: number): Float32Array { export function zeros(length: number): Float32Array {
return new Float32Array(length) return new Float32Array(length)
} }
...@@ -21,3 +24,84 @@ export async function expectAllTensorsReleased(fn: () => any) { ...@@ -21,3 +24,84 @@ export async function expectAllTensorsReleased(fn: () => any) {
export function tensor3D() { export function tensor3D() {
return tf.tensor3d([[[0]]]) return tf.tensor3d([[[0]]])
} }
export type WithNetOptions = {
quantized?: boolean
}
export type InjectNetArgs = {
faceDetectionNet: faceapi.FaceDetectionNet
faceLandmarkNet: faceapi.FaceLandmarkNet
faceRecognitionNet: faceapi.FaceRecognitionNet
}
export type DescribeWithNetsOptions = {
withFaceDetectionNet?: WithNetOptions
withFaceLandmarkNet?: WithNetOptions
withFaceRecognitionNet?: WithNetOptions
}
async function loadNetWeights(uri: string): Promise<Float32Array> {
return new Float32Array(await (await fetch(uri)).arrayBuffer())
}
async function initNet<TNet extends NeuralNetwork<any>>(
net: TNet,
uncompressedFilename: string | boolean
) {
await net.load(
uncompressedFilename
? await loadNetWeights(`base/weights/uncompressed/${uncompressedFilename}`)
: 'base/weights'
)
}
export function describeWithNets(
description: string,
options: DescribeWithNetsOptions,
specDefinitions: (nets: InjectNetArgs) => void
) {
describe(description, () => {
let faceDetectionNet: faceapi.FaceDetectionNet = new faceapi.FaceDetectionNet()
let faceLandmarkNet: faceapi.FaceLandmarkNet = new faceapi.FaceLandmarkNet()
let faceRecognitionNet: faceapi.FaceRecognitionNet = new faceapi.FaceRecognitionNet()
beforeAll(async () => {
const {
withFaceDetectionNet,
withFaceLandmarkNet,
withFaceRecognitionNet
} = options
if (withFaceDetectionNet) {
await initNet<faceapi.FaceDetectionNet>(
faceDetectionNet,
!withFaceDetectionNet.quantized && 'face_detection_model.weights'
)
}
if (withFaceLandmarkNet) {
await initNet<faceapi.FaceLandmarkNet>(
faceLandmarkNet,
!withFaceLandmarkNet.quantized && 'face_landmark_68_model.weights'
)
}
if (withFaceRecognitionNet) {
await initNet<faceapi.FaceRecognitionNet>(
faceRecognitionNet,
!withFaceRecognitionNet.quantized && 'face_recognition_model.weights'
)
}
})
afterAll(() => {
faceDetectionNet && faceDetectionNet.dispose()
faceLandmarkNet && faceLandmarkNet.dispose()
faceRecognitionNet && faceRecognitionNet.dispose()
})
specDefinitions({ faceDetectionNet, faceLandmarkNet, faceRecognitionNet })
})
}
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment