Skip to content

F
face
Unverified Commit a6a68a56 by justadudewhohacks Committed by GitHub
Options

Merge pull request #40 from justadudewhohacks/trainable

NeuralNetwork base class + trainable
parents 57b51bd4 4b69fc27
Showing with 990 additions and 157 deletions
build/commons/NeuralNetwork.d.ts 0 → 100644
import * as tf from '@tensorflow/tfjs-core';
import { ParamMapping } from './types';
export declare class NeuralNetwork<TNetParams> {
protected _params: TNetParams | undefined;
protected _paramMappings: ParamMapping[];
readonly params: TNetParams | undefined;
readonly paramMappings: ParamMapping[];
getParamFromPath(paramPath: string): tf.Tensor;
reassignParamFromPath(paramPath: string, tensor: tf.Tensor): void;
getParamList(): {
path: string;
tensor: tf.Tensor<tf.Rank>;
}[];
getTrainableParams(): {
path: string;
tensor: tf.Tensor<tf.Rank>;
}[];
getFrozenParams(): {
path: string;
tensor: tf.Tensor<tf.Rank>;
}[];
variable(): void;
freeze(): void;
dispose(): void;
private traversePropertyPath(paramPath);
}
build/commons/NeuralNetwork.js 0 → 100644
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
var tf = require("@tensorflow/tfjs-core");
var NeuralNetwork = /** @class */ (function () {
function NeuralNetwork() {
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) {
var _a = this.traversePropertyPath(paramPath), obj = _a.obj, objProp = _a.objProp;
obj[objProp].dispose();
obj[objProp] = tensor;
};
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 tf.Variable; });
};
NeuralNetwork.prototype.getFrozenParams = function () {
return this.getParamList().filter(function (param) { return !(param.tensor instanceof tf.Variable); });
};
NeuralNetwork.prototype.variable = function () {
var _this = this;
this.getFrozenParams().forEach(function (_a) {
var path = _a.path, tensor = _a.tensor;
_this.reassignParamFromPath(path, tf.variable(tensor));
});
};
NeuralNetwork.prototype.freeze = function () {
var _this = this;
this.getTrainableParams().forEach(function (_a) {
var path = _a.path, tensor = _a.tensor;
_this.reassignParamFromPath(path, tf.tensor(tensor));
});
};
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 tf.Tensor)) {
throw new Error("traversePropertyPath - parameter is not a tensor, for path " + paramPath);
}
return { obj: obj, objProp: objProp };
};
return NeuralNetwork;
}());
exports.NeuralNetwork = NeuralNetwork;
//# sourceMappingURL=NeuralNetwork.js.map
\ No newline at end of file
build/commons/NeuralNetwork.js.map 0 → 100644
{"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"}
\ No newline at end of file
build/commons/extractConvParamsFactory.d.ts
import { ConvParams, ExtractWeightsFunction } from './types';
export declare function extractConvParamsFactory(extractWeights: ExtractWeightsFunction): (channelsIn: number, channelsOut: number, filterSize: number) => ConvParams;
import { ConvParams, ExtractWeightsFunction, ParamMapping } from './types';
export declare function extractConvParamsFactory(extractWeights: ExtractWeightsFunction, paramMappings: ParamMapping[]): (channelsIn: number, channelsOut: number, filterSize: number, mappedPrefix: string) => ConvParams;
build/commons/extractConvParamsFactory.js
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
var tf = require("@tensorflow/tfjs-core");
function extractConvParamsFactory(extractWeights) {
return function (channelsIn, channelsOut, filterSize) {
function extractConvParamsFactory(extractWeights, paramMappings) {
return function (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
......
build/commons/extractConvParamsFactory.js.map
{"version":3,"file":"extractConvParamsFactory.js","sourceRoot":"","sources":["../../src/commons/extractConvParamsFactory.ts"],"names":[],"mappings":";;AAAA,0CAA4C;AAI5C,kCAAyC,cAAsC;IAC7E,OAAO,UACL,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,CAAA;AACH,CAAC;AAjBD,4DAiBC"}
\ No newline at end of file
{"version":3,"file":"extractConvParamsFactory.js","sourceRoot":"","sources":["../../src/commons/extractConvParamsFactory.ts"],"names":[],"mappings":";;AAAA,0CAA4C;AAI5C,kCAAyC,cAAsC,EAAE,aAA6B;IAC5G,OAAO,UACL,UAAkB,EAClB,WAAmB,EACnB,UAAkB,EAClB,YAAoB;QAEpB,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;YACL,OAAO,SAAA;YACP,IAAI,MAAA;SACL,CAAA;IACH,CAAC,CAAA;AACH,CAAC;AAvBD,4DAuBC"}
\ No newline at end of file
build/commons/extractWeightEntry.d.ts 0 → 100644
export declare function extractWeightEntry(weightMap: any, path: string, paramRank: number): {
path: string;
tensor: any;
};
build/commons/extractWeightEntry.js 0 → 100644
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
var isTensor_1 = require("./isTensor");
function extractWeightEntry(weightMap, path, paramRank) {
var tensor = weightMap[path];
if (!isTensor_1.isTensor(tensor, paramRank)) {
throw new Error("expected weightMap[" + path + "] to be a Tensor" + paramRank + "D, instead have " + tensor);
}
return { path: path, tensor: tensor };
}
exports.extractWeightEntry = extractWeightEntry;
//# sourceMappingURL=extractWeightEntry.js.map
\ No newline at end of file
build/commons/extractWeightEntry.js.map 0 → 100644
{"version":3,"file":"extractWeightEntry.js","sourceRoot":"","sources":["../../src/commons/extractWeightEntry.ts"],"names":[],"mappings":";;AAAA,uCAAsC;AAEtC,4BAAmC,SAAc,EAAE,IAAY,EAAE,SAAiB;IAChF,IAAM,MAAM,GAAG,SAAS,CAAC,IAAI,CAAC,CAAA;IAE9B,IAAI,CAAC,mBAAQ,CAAC,MAAM,EAAE,SAAS,CAAC,EAAE;QAChC,MAAM,IAAI,KAAK,CAAC,wBAAsB,IAAI,wBAAmB,SAAS,wBAAmB,MAAQ,CAAC,CAAA;KACnG;IAED,OAAO,EAAE,IAAI,MAAA,EAAE,MAAM,QAAA,EAAE,CAAA;AACzB,CAAC;AARD,gDAQC"}
\ No newline at end of file
build/commons/types.d.ts
......@@ -10,3 +10,7 @@ export declare type BatchReshapeInfo = {
paddingX: number;
paddingY: number;
};
export declare type ParamMapping = {
originalPath?: string;
paramPath: string;
};
build/faceLandmarkNet/FaceLandmarkNet.d.ts
import * as tf from '@tensorflow/tfjs-core';
import { NeuralNetwork } from '../commons/NeuralNetwork';
import { NetInput } from '../NetInput';
import { TNetInput } from '../types';
import { FaceLandmarks } from './FaceLandmarks';
export declare class FaceLandmarkNet {
private _params;
import { NetParams } from './types';
export declare class FaceLandmarkNet extends NeuralNetwork<NetParams> {
load(weightsOrUrl: Float32Array | string | undefined): Promise<void>;
extractWeights(weights: Float32Array): void;
forwardInput(input: NetInput): tf.Tensor2D;
......
build/faceLandmarkNet/FaceLandmarkNet.js
......@@ -3,6 +3,7 @@ Object.defineProperty(exports, "__esModule", { value: true });
var tslib_1 = require("tslib");
var tf = require("@tensorflow/tfjs-core");
var convLayer_1 = require("../commons/convLayer");
var NeuralNetwork_1 = require("../commons/NeuralNetwork");
var Point_1 = require("../Point");
var toNetInput_1 = require("../toNetInput");
var utils_1 = require("../utils");
......@@ -17,12 +18,14 @@ function maxPool(x, strides) {
if (strides === void 0) { strides = [2, 2]; }
return tf.maxPool(x, [2, 2], strides, 'valid');
}
var FaceLandmarkNet = /** @class */ (function () {
var FaceLandmarkNet = /** @class */ (function (_super) {
tslib_1.__extends(FaceLandmarkNet, _super);
function FaceLandmarkNet() {
return _super !== null && _super.apply(this, arguments) || this;
}
FaceLandmarkNet.prototype.load = function (weightsOrUrl) {
return tslib_1.__awaiter(this, void 0, void 0, function () {
var _a;
var _a, paramMappings, params;
return tslib_1.__generator(this, function (_b) {
switch (_b.label) {
case 0:
......@@ -33,17 +36,20 @@ var FaceLandmarkNet = /** @class */ (function () {
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();
_a = _b.sent(), paramMappings = _a.paramMappings, params = _a.params;
this._paramMappings = paramMappings;
this._params = params;
return [2 /*return*/];
}
});
});
};
FaceLandmarkNet.prototype.extractWeights = function (weights) {
this._params = extractParams_1.extractParams(weights);
var _a = extractParams_1.extractParams(weights), paramMappings = _a.paramMappings, params = _a.params;
this._paramMappings = paramMappings;
this._params = params;
};
FaceLandmarkNet.prototype.forwardInput = function (input) {
var params = this._params;
......@@ -141,6 +147,6 @@ var FaceLandmarkNet = /** @class */ (function () {
});
};
return FaceLandmarkNet;
}());
}(NeuralNetwork_1.NeuralNetwork));
exports.FaceLandmarkNet = FaceLandmarkNet;
//# sourceMappingURL=FaceLandmarkNet.js.map
\ No newline at end of file
build/faceLandmarkNet/FaceLandmarkNet.js.map
{"version":3,"file":"FaceLandmarkNet.js","sourceRoot":"","sources":["../../src/faceLandmarkNet/FaceLandmarkNet.ts"],"names":[],"mappings":";;;AAAA,0CAA4C;AAE5C,kDAAiD;AAGjD,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;IAAA;IAiHA,CAAC;IA7Gc,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;wBACD,KAAA,IAAI,CAAA;wBAAW,qBAAM,yCAAmB,CAAC,YAAY,CAAC,EAAA;;wBAAtD,GAAK,OAAO,GAAG,SAAuC,CAAA;;;;;KACvD;IAEM,wCAAc,GAArB,UAAsB,OAAqB;QACzC,IAAI,CAAC,OAAO,GAAG,6BAAa,CAAC,OAAO,CAAC,CAAA;IACvC,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,AAjHD,IAiHC;AAjHY,0CAAe"}
\ No newline at end of file
{"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"}
\ No newline at end of file
build/faceLandmarkNet/extractParams.d.ts
import { ParamMapping } from '../commons/types';
import { NetParams } from './types';
export declare function extractParams(weights: Float32Array): NetParams;
export declare function extractParams(weights: Float32Array): {
params: NetParams;
paramMappings: ParamMapping[];
};
build/faceLandmarkNet/extractParams.js
......@@ -4,40 +4,45 @@ var tf = require("@tensorflow/tfjs-core");
var extractConvParamsFactory_1 = require("../commons/extractConvParamsFactory");
var extractWeightsFactory_1 = require("../commons/extractWeightsFactory");
function extractParams(weights) {
var paramMappings = [];
var _a = extractWeightsFactory_1.extractWeightsFactory(weights), extractWeights = _a.extractWeights, getRemainingWeights = _a.getRemainingWeights;
var extractConvParams = extractConvParamsFactory_1.extractConvParamsFactory(extractWeights);
function extractFcParams(channelsIn, channelsOut) {
var extractConvParams = extractConvParamsFactory_1.extractConvParamsFactory(extractWeights, paramMappings);
function extractFcParams(channelsIn, channelsOut, mappedPrefix) {
var fc_weights = tf.tensor2d(extractWeights(channelsIn * channelsOut), [channelsIn, channelsOut]);
var fc_bias = tf.tensor1d(extractWeights(channelsOut));
paramMappings.push({ paramPath: mappedPrefix + "/weights" }, { paramPath: mappedPrefix + "/bias" });
return {
weights: fc_weights,
bias: fc_bias
};
}
var conv0_params = extractConvParams(3, 32, 3);
var conv1_params = extractConvParams(32, 64, 3);
var conv2_params = extractConvParams(64, 64, 3);
var conv3_params = extractConvParams(64, 64, 3);
var conv4_params = extractConvParams(64, 64, 3);
var conv5_params = extractConvParams(64, 128, 3);
var conv6_params = extractConvParams(128, 128, 3);
var conv7_params = extractConvParams(128, 256, 3);
var fc0_params = extractFcParams(6400, 1024);
var fc1_params = extractFcParams(1024, 136);
var conv0_params = extractConvParams(3, 32, 3, 'conv0_params');
var conv1_params = extractConvParams(32, 64, 3, 'conv1_params');
var conv2_params = extractConvParams(64, 64, 3, 'conv2_params');
var conv3_params = extractConvParams(64, 64, 3, 'conv3_params');
var conv4_params = extractConvParams(64, 64, 3, 'conv4_params');
var conv5_params = extractConvParams(64, 128, 3, 'conv5_params');
var conv6_params = extractConvParams(128, 128, 3, 'conv6_params');
var conv7_params = extractConvParams(128, 256, 3, 'conv7_params');
var fc0_params = extractFcParams(6400, 1024, 'fc0_params');
var fc1_params = extractFcParams(1024, 136, 'fc1_params');
if (getRemainingWeights().length !== 0) {
throw new Error("weights remaing after extract: " + getRemainingWeights().length);
}
return {
conv0_params: conv0_params,
conv1_params: conv1_params,
conv2_params: conv2_params,
conv3_params: conv3_params,
conv4_params: conv4_params,
conv5_params: conv5_params,
conv6_params: conv6_params,
conv7_params: conv7_params,
fc0_params: fc0_params,
fc1_params: fc1_params
paramMappings: paramMappings,
params: {
conv0_params: conv0_params,
conv1_params: conv1_params,
conv2_params: conv2_params,
conv3_params: conv3_params,
conv4_params: conv4_params,
conv5_params: conv5_params,
conv6_params: conv6_params,
conv7_params: conv7_params,
fc0_params: fc0_params,
fc1_params: fc1_params
}
};
}
exports.extractParams = extractParams;
......
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;AAGzE,uBAA8B,OAAqB;IAC3C,IAAA,2DAG4B,EAFhC,kCAAc,EACd,4CAAmB,CACa;IAElC,IAAM,iBAAiB,GAAG,mDAAwB,CAAC,cAAc,CAAC,CAAA;IAElE,yBAAyB,UAAkB,EAAE,WAAmB;QAC9D,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;QACxD,OAAO;YACL,OAAO,EAAE,UAAU;YACnB,IAAI,EAAE,OAAO;SACd,CAAA;IACH,CAAC;IAED,IAAM,YAAY,GAAG,iBAAiB,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;IAChD,IAAM,YAAY,GAAG,iBAAiB,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;IACjD,IAAM,YAAY,GAAG,iBAAiB,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;IACjD,IAAM,YAAY,GAAG,iBAAiB,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;IACjD,IAAM,YAAY,GAAG,iBAAiB,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC,CAAC,CAAA;IACjD,IAAM,YAAY,GAAG,iBAAiB,CAAC,EAAE,EAAE,GAAG,EAAE,CAAC,CAAC,CAAA;IAClD,IAAM,YAAY,GAAG,iBAAiB,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC,CAAC,CAAA;IACnD,IAAM,YAAY,GAAG,iBAAiB,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC,CAAC,CAAA;IACnD,IAAM,UAAU,GAAG,eAAe,CAAC,IAAI,EAAE,IAAI,CAAC,CAAA;IAC9C,IAAM,UAAU,GAAG,eAAe,CAAC,IAAI,EAAE,GAAG,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,YAAY,cAAA;QACZ,YAAY,cAAA;QACZ,YAAY,cAAA;QACZ,YAAY,cAAA;QACZ,YAAY,cAAA;QACZ,YAAY,cAAA;QACZ,YAAY,cAAA;QACZ,YAAY,cAAA;QACZ,UAAU,YAAA;QACV,UAAU,YAAA;KACX,CAAA;AACH,CAAC;AA5CD,sCA4CC"}
\ No newline at end of file
{"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"}
\ No newline at end of file
build/faceLandmarkNet/loadQuantizedParams.d.ts
import { ParamMapping } from '../commons/types';
import { NetParams } from './types';
export declare function loadQuantizedParams(uri: string | undefined): Promise<NetParams>;
export declare function loadQuantizedParams(uri: string | undefined): Promise<{
params: NetParams;
paramMappings: ParamMapping[];
}>;
build/faceLandmarkNet/loadQuantizedParams.js
"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
var tslib_1 = require("tslib");
var extractWeightEntry_1 = require("../commons/extractWeightEntry");
var loadWeightMap_1 = require("../commons/loadWeightMap");
var isTensor_1 = require("../commons/isTensor");
var DEFAULT_MODEL_NAME = 'face_landmark_68_model';
function extractorsFactory(weightMap) {
function extractConvParams(prefix) {
var params = {
filters: weightMap[prefix + "/kernel"],
bias: weightMap[prefix + "/bias"]
function extractorsFactory(weightMap, paramMappings) {
function extractConvParams(prefix, mappedPrefix) {
var filtersEntry = extractWeightEntry_1.extractWeightEntry(weightMap, prefix + "/kernel", 4);
var biasEntry = extractWeightEntry_1.extractWeightEntry(weightMap, prefix + "/bias", 1);
paramMappings.push({ originalPath: filtersEntry.path, paramPath: mappedPrefix + "/filters" }, { originalPath: biasEntry.path, paramPath: mappedPrefix + "/bias" });
return {
filters: filtersEntry.tensor,
bias: biasEntry.tensor
};
if (!isTensor_1.isTensor4D(params.filters)) {
throw new Error("expected weightMap[" + prefix + "/kernel] to be a Tensor4D, instead have " + params.filters);
}
if (!isTensor_1.isTensor1D(params.bias)) {
throw new Error("expected weightMap[" + prefix + "/bias] to be a Tensor1D, instead have " + params.bias);
}
return params;
}
function extractFcParams(prefix) {
var params = {
weights: weightMap[prefix + "/kernel"],
bias: weightMap[prefix + "/bias"]
function extractFcParams(prefix, mappedPrefix) {
var weightsEntry = extractWeightEntry_1.extractWeightEntry(weightMap, prefix + "/kernel", 2);
var biasEntry = extractWeightEntry_1.extractWeightEntry(weightMap, prefix + "/bias", 1);
paramMappings.push({ originalPath: weightsEntry.path, paramPath: mappedPrefix + "/weights" }, { originalPath: biasEntry.path, paramPath: mappedPrefix + "/bias" });
return {
weights: weightsEntry.tensor,
bias: biasEntry.tensor
};
if (!isTensor_1.isTensor2D(params.weights)) {
throw new Error("expected weightMap[" + prefix + "/kernel] to be a Tensor2D, instead have " + params.weights);
}
if (!isTensor_1.isTensor1D(params.bias)) {
throw new Error("expected weightMap[" + prefix + "/bias] to be a Tensor1D, instead have " + params.bias);
}
return params;
}
return {
extractConvParams: extractConvParams,
......@@ -38,25 +30,27 @@ function extractorsFactory(weightMap) {
}
function loadQuantizedParams(uri) {
return tslib_1.__awaiter(this, void 0, void 0, function () {
var weightMap, _a, extractConvParams, extractFcParams;
var weightMap, paramMappings, _a, extractConvParams, extractFcParams, params;
return tslib_1.__generator(this, function (_b) {
switch (_b.label) {
case 0: return [4 /*yield*/, loadWeightMap_1.loadWeightMap(uri, DEFAULT_MODEL_NAME)];
case 1:
weightMap = _b.sent();
_a = extractorsFactory(weightMap), extractConvParams = _a.extractConvParams, extractFcParams = _a.extractFcParams;
return [2 /*return*/, {
conv0_params: extractConvParams('conv2d_0'),
conv1_params: extractConvParams('conv2d_1'),
conv2_params: extractConvParams('conv2d_2'),
conv3_params: extractConvParams('conv2d_3'),
conv4_params: extractConvParams('conv2d_4'),
conv5_params: extractConvParams('conv2d_5'),
conv6_params: extractConvParams('conv2d_6'),
conv7_params: extractConvParams('conv2d_7'),
fc0_params: extractFcParams('dense'),
fc1_params: extractFcParams('logits')
}];
paramMappings = [];
_a = extractorsFactory(weightMap, paramMappings), extractConvParams = _a.extractConvParams, extractFcParams = _a.extractFcParams;
params = {
conv0_params: extractConvParams('conv2d_0', 'conv0_params'),
conv1_params: extractConvParams('conv2d_1', 'conv1_params'),
conv2_params: extractConvParams('conv2d_2', 'conv2_params'),
conv3_params: extractConvParams('conv2d_3', 'conv3_params'),
conv4_params: extractConvParams('conv2d_4', 'conv4_params'),
conv5_params: extractConvParams('conv2d_5', 'conv5_params'),
conv6_params: extractConvParams('conv2d_6', 'conv6_params'),
conv7_params: extractConvParams('conv2d_7', 'conv7_params'),
fc0_params: extractFcParams('dense', 'fc0_params'),
fc1_params: extractFcParams('logits', 'fc1_params')
};
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,0DAAyD;AAGzD,gDAAyE;AAEzE,IAAM,kBAAkB,GAAG,wBAAwB,CAAA;AAEnD,2BAA2B,SAAc;IAEvC,2BAA2B,MAAc;QACvC,IAAM,MAAM,GAAG;YACb,OAAO,EAAE,SAAS,CAAI,MAAM,YAAS,CAAgB;YACrD,IAAI,EAAE,SAAS,CAAI,MAAM,UAAO,CAAgB;SACjD,CAAA;QAED,IAAI,CAAC,qBAAU,CAAC,MAAM,CAAC,OAAO,CAAC,EAAE;YAC/B,MAAM,IAAI,KAAK,CAAC,wBAAsB,MAAM,gDAA2C,MAAM,CAAC,OAAS,CAAC,CAAA;SACzG;QAED,IAAI,CAAC,qBAAU,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE;YAC5B,MAAM,IAAI,KAAK,CAAC,wBAAsB,MAAM,8CAAyC,MAAM,CAAC,IAAM,CAAC,CAAA;SACpG;QAED,OAAO,MAAM,CAAA;IACf,CAAC;IAED,yBAAyB,MAAc;QACrC,IAAM,MAAM,GAAG;YACb,OAAO,EAAE,SAAS,CAAI,MAAM,YAAS,CAAgB;YACrD,IAAI,EAAE,SAAS,CAAI,MAAM,UAAO,CAAgB;SACjD,CAAA;QAED,IAAI,CAAC,qBAAU,CAAC,MAAM,CAAC,OAAO,CAAC,EAAE;YAC/B,MAAM,IAAI,KAAK,CAAC,wBAAsB,MAAM,gDAA2C,MAAM,CAAC,OAAS,CAAC,CAAA;SACzG;QAED,IAAI,CAAC,qBAAU,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE;YAC5B,MAAM,IAAI,KAAK,CAAC,wBAAsB,MAAM,8CAAyC,MAAM,CAAC,IAAM,CAAC,CAAA;SACpG;QAED,OAAO,MAAM,CAAA;IACf,CAAC;IAED,OAAO;QACL,iBAAiB,mBAAA;QACjB,eAAe,iBAAA;KAChB,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,iBAAiB,uBAAA,EACjB,eAAe,qBAAA,CACe;oBAEhC,sBAAO;4BACL,YAAY,EAAE,iBAAiB,CAAC,UAAU,CAAC;4BAC3C,YAAY,EAAE,iBAAiB,CAAC,UAAU,CAAC;4BAC3C,YAAY,EAAE,iBAAiB,CAAC,UAAU,CAAC;4BAC3C,YAAY,EAAE,iBAAiB,CAAC,UAAU,CAAC;4BAC3C,YAAY,EAAE,iBAAiB,CAAC,UAAU,CAAC;4BAC3C,YAAY,EAAE,iBAAiB,CAAC,UAAU,CAAC;4BAC3C,YAAY,EAAE,iBAAiB,CAAC,UAAU,CAAC;4BAC3C,YAAY,EAAE,iBAAiB,CAAC,UAAU,CAAC;4BAC3C,UAAU,EAAE,eAAe,CAAC,OAAO,CAAC;4BACpC,UAAU,EAAE,eAAe,CAAC,QAAQ,CAAC;yBACtC,EAAA;;;;CACF;AApBD,kDAoBC"}
\ No newline at end of file
{"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"}
\ No newline at end of file
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/commons/NeuralNetwork.ts 0 → 100644
import * as tf from '@tensorflow/tfjs-core';
import { ParamMapping } from './types';
export class NeuralNetwork<TNetParams> {
protected _params: TNetParams | undefined = undefined
protected _paramMappings: ParamMapping[] = []
public get params(): TNetParams | undefined {
return this._params
}
public get paramMappings(): ParamMapping[] {
return this._paramMappings
}
public getParamFromPath(paramPath: string): tf.Tensor {
const { obj, objProp } = this.traversePropertyPath(paramPath)
return obj[objProp]
}
public reassignParamFromPath(paramPath: string, tensor: tf.Tensor) {
const { obj, objProp } = this.traversePropertyPath(paramPath)
obj[objProp].dispose()
obj[objProp] = tensor
}
public getParamList() {
return this._paramMappings.map(({ paramPath }) => ({
path: paramPath,
tensor: this.getParamFromPath(paramPath)
}))
}
public getTrainableParams() {
return this.getParamList().filter(param => param.tensor instanceof tf.Variable)
}
public getFrozenParams() {
return this.getParamList().filter(param => !(param.tensor instanceof tf.Variable))
}
public variable() {
this.getFrozenParams().forEach(({ path, tensor }) => {
this.reassignParamFromPath(path, tf.variable(tensor))
})
}
public freeze() {
this.getTrainableParams().forEach(({ path, tensor }) => {
this.reassignParamFromPath(path, tf.tensor(tensor as any))
})
}
public dispose() {
this.getParamList().forEach(param => param.tensor.dispose())
this._params = undefined
}
private traversePropertyPath(paramPath: string) {
if (!this.params) {
throw new Error(`traversePropertyPath - model has no loaded params`)
}
const result = paramPath.split('/').reduce((res: { nextObj: any, obj?: any, objProp?: string }, objProp) => {
if (!res.nextObj.hasOwnProperty(objProp)) {
throw new Error(`traversePropertyPath - object does not have property ${objProp}, for path ${paramPath}`)
}
return { obj: res.nextObj, objProp, nextObj: res.nextObj[objProp] }
}, { nextObj: this.params })
const { obj, objProp } = result
if (!obj || !objProp || !(obj[objProp] instanceof tf.Tensor)) {
throw new Error(`traversePropertyPath - parameter is not a tensor, for path ${paramPath}`)
}
return { obj, objProp }
}
}
\ No newline at end of file
src/commons/extractConvParamsFactory.ts
import * as tf from '@tensorflow/tfjs-core';
import { ConvParams, ExtractWeightsFunction } from './types';
import { ConvParams, ExtractWeightsFunction, ParamMapping } from './types';
export function extractConvParamsFactory(extractWeights: ExtractWeightsFunction) {
export function extractConvParamsFactory(extractWeights: ExtractWeightsFunction, paramMappings: ParamMapping[]) {
return function (
channelsIn: number,
channelsOut: number,
filterSize: number
filterSize: number,
mappedPrefix: string
): ConvParams {
const filters = tf.tensor4d(
extractWeights(channelsIn * channelsOut * filterSize * filterSize),
......@@ -14,6 +15,11 @@ export function extractConvParamsFactory(extractWeights: ExtractWeightsFunction)
)
const bias = tf.tensor1d(extractWeights(channelsOut))
paramMappings.push(
{ paramPath: `${mappedPrefix}/filters` },
{ paramPath: `${mappedPrefix}/bias` }
)
return {
filters,
bias
......
src/commons/extractWeightEntry.ts 0 → 100644
import { isTensor } from './isTensor';
export function extractWeightEntry(weightMap: any, path: string, paramRank: number) {
const tensor = weightMap[path]
if (!isTensor(tensor, paramRank)) {
throw new Error(`expected weightMap[${path}] to be a Tensor${paramRank}D, instead have ${tensor}`)
}
return { path, tensor }
}
\ No newline at end of file
src/commons/types.ts
......@@ -13,3 +13,8 @@ export type BatchReshapeInfo = {
paddingX: number
paddingY: number
}
export type ParamMapping = {
originalPath?: string
paramPath: string
}
\ No newline at end of file
src/faceLandmarkNet/FaceLandmarkNet.ts
import * as tf from '@tensorflow/tfjs-core';
import { convLayer } from '../commons/convLayer';
import { NeuralNetwork } from '../commons/NeuralNetwork';
import { ConvParams } from '../commons/types';
import { NetInput } from '../NetInput';
import { Point } from '../Point';
......@@ -21,9 +22,7 @@ function maxPool(x: tf.Tensor4D, strides: [number, number] = [2, 2]): tf.Tensor4
return tf.maxPool(x, [2, 2], strides, 'valid')
}
export class FaceLandmarkNet {
private _params: NetParams
export class FaceLandmarkNet extends NeuralNetwork<NetParams> {
public async load(weightsOrUrl: Float32Array | string | undefined): Promise<void> {
if (weightsOrUrl instanceof Float32Array) {
......@@ -34,11 +33,23 @@ export class FaceLandmarkNet {
if (weightsOrUrl && typeof weightsOrUrl !== 'string') {
throw new Error('FaceLandmarkNet.load - expected model uri, or weights as Float32Array')
}
this._params = await loadQuantizedParams(weightsOrUrl)
const {
paramMappings,
params
} = await loadQuantizedParams(weightsOrUrl)
this._paramMappings = paramMappings
this._params = params
}
public extractWeights(weights: Float32Array) {
this._params = extractParams(weights)
const {
paramMappings,
params
} = extractParams(weights)
this._paramMappings = paramMappings
this._params = params
}
public forwardInput(input: NetInput): tf.Tensor2D {
......
src/faceLandmarkNet/extractParams.ts
......@@ -2,50 +2,62 @@ import * as tf from '@tensorflow/tfjs-core';
import { extractConvParamsFactory } from '../commons/extractConvParamsFactory';
import { extractWeightsFactory } from '../commons/extractWeightsFactory';
import { ParamMapping } from '../commons/types';
import { FCParams, NetParams } from './types';
export function extractParams(weights: Float32Array): NetParams {
export function extractParams(weights: Float32Array): { params: NetParams, paramMappings: ParamMapping[] } {
const paramMappings: ParamMapping[] = []
const {
extractWeights,
getRemainingWeights
} = extractWeightsFactory(weights)
const extractConvParams = extractConvParamsFactory(extractWeights)
const extractConvParams = extractConvParamsFactory(extractWeights, paramMappings)
function extractFcParams(channelsIn: number, channelsOut: number,): FCParams {
function extractFcParams(channelsIn: number, channelsOut: number, mappedPrefix: string): FCParams {
const fc_weights = tf.tensor2d(extractWeights(channelsIn * channelsOut), [channelsIn, channelsOut])
const fc_bias = tf.tensor1d(extractWeights(channelsOut))
paramMappings.push(
{ paramPath: `${mappedPrefix}/weights` },
{ paramPath: `${mappedPrefix}/bias` }
)
return {
weights: fc_weights,
bias: fc_bias
}
}
const conv0_params = extractConvParams(3, 32, 3)
const conv1_params = extractConvParams(32, 64, 3)
const conv2_params = extractConvParams(64, 64, 3)
const conv3_params = extractConvParams(64, 64, 3)
const conv4_params = extractConvParams(64, 64, 3)
const conv5_params = extractConvParams(64, 128, 3)
const conv6_params = extractConvParams(128, 128, 3)
const conv7_params = extractConvParams(128, 256, 3)
const fc0_params = extractFcParams(6400, 1024)
const fc1_params = extractFcParams(1024, 136)
const conv0_params = extractConvParams(3, 32, 3, 'conv0_params')
const conv1_params = extractConvParams(32, 64, 3, 'conv1_params')
const conv2_params = extractConvParams(64, 64, 3, 'conv2_params')
const conv3_params = extractConvParams(64, 64, 3, 'conv3_params')
const conv4_params = extractConvParams(64, 64, 3, 'conv4_params')
const conv5_params = extractConvParams(64, 128, 3, 'conv5_params')
const conv6_params = extractConvParams(128, 128, 3, 'conv6_params')
const conv7_params = extractConvParams(128, 256, 3, 'conv7_params')
const fc0_params = extractFcParams(6400, 1024, 'fc0_params')
const fc1_params = extractFcParams(1024, 136, 'fc1_params')
if (getRemainingWeights().length !== 0) {
throw new Error(`weights remaing after extract: ${getRemainingWeights().length}`)
}
return {
conv0_params,
conv1_params,
conv2_params,
conv3_params,
conv4_params,
conv5_params,
conv6_params,
conv7_params,
fc0_params,
fc1_params
paramMappings,
params: {
conv0_params,
conv1_params,
conv2_params,
conv3_params,
conv4_params,
conv5_params,
conv6_params,
conv7_params,
fc0_params,
fc1_params
}
}
}
\ No newline at end of file
src/faceLandmarkNet/loadQuantizedParams.ts
import * as tf from '@tensorflow/tfjs-core';
import { extractWeightEntry } from '../commons/extractWeightEntry';
import { loadWeightMap } from '../commons/loadWeightMap';
import { ConvParams } from '../commons/types';
import { ConvParams, ParamMapping } from '../commons/types';
import { FCParams, NetParams } from './types';
import { isTensor4D, isTensor1D, isTensor2D } from '../commons/isTensor';
const DEFAULT_MODEL_NAME = 'face_landmark_68_model'
function extractorsFactory(weightMap: any) {
function extractorsFactory(weightMap: any, paramMappings: ParamMapping[]) {
function extractConvParams(prefix: string): ConvParams {
const params = {
filters: weightMap[`${prefix}/kernel`] as tf.Tensor4D,
bias: weightMap[`${prefix}/bias`] as tf.Tensor1D
function extractConvParams(prefix: string, mappedPrefix: string): ConvParams {
const filtersEntry = extractWeightEntry(weightMap, `${prefix}/kernel`, 4)
const biasEntry = extractWeightEntry(weightMap, `${prefix}/bias`, 1)
paramMappings.push(
{ originalPath: filtersEntry.path, paramPath: `${mappedPrefix}/filters` },
{ originalPath: biasEntry.path, paramPath: `${mappedPrefix}/bias` }
)
return {
filters: filtersEntry.tensor as tf.Tensor4D,
bias: biasEntry.tensor as tf.Tensor1D
}
if (!isTensor4D(params.filters)) {
throw new Error(`expected weightMap[${prefix}/kernel] to be a Tensor4D, instead have ${params.filters}`)
}
if (!isTensor1D(params.bias)) {
throw new Error(`expected weightMap[${prefix}/bias] to be a Tensor1D, instead have ${params.bias}`)
}
return params
}
function extractFcParams(prefix: string): FCParams {
const params = {
weights: weightMap[`${prefix}/kernel`] as tf.Tensor2D,
bias: weightMap[`${prefix}/bias`] as tf.Tensor1D
}
if (!isTensor2D(params.weights)) {
throw new Error(`expected weightMap[${prefix}/kernel] to be a Tensor2D, instead have ${params.weights}`)
function extractFcParams(prefix: string, mappedPrefix: string): FCParams {
const weightsEntry = extractWeightEntry(weightMap, `${prefix}/kernel`, 2)
const biasEntry = extractWeightEntry(weightMap, `${prefix}/bias`, 1)
paramMappings.push(
{ originalPath: weightsEntry.path, paramPath: `${mappedPrefix}/weights` },
{ originalPath: biasEntry.path, paramPath: `${mappedPrefix}/bias` }
)
return {
weights: weightsEntry.tensor as tf.Tensor2D,
bias: biasEntry.tensor as tf.Tensor1D
}
if (!isTensor1D(params.bias)) {
throw new Error(`expected weightMap[${prefix}/bias] to be a Tensor1D, instead have ${params.bias}`)
}
return params
}
return {
......@@ -49,24 +41,30 @@ function extractorsFactory(weightMap: any) {
}
}
export async function loadQuantizedParams(uri: string | undefined): Promise<NetParams> {
export async function loadQuantizedParams(
uri: string | undefined
): Promise<{ params: NetParams, paramMappings: ParamMapping[] }> {
const weightMap = await loadWeightMap(uri, DEFAULT_MODEL_NAME)
const paramMappings: ParamMapping[] = []
const {
extractConvParams,
extractFcParams
} = extractorsFactory(weightMap)
} = extractorsFactory(weightMap, paramMappings)
return {
conv0_params: extractConvParams('conv2d_0'),
conv1_params: extractConvParams('conv2d_1'),
conv2_params: extractConvParams('conv2d_2'),
conv3_params: extractConvParams('conv2d_3'),
conv4_params: extractConvParams('conv2d_4'),
conv5_params: extractConvParams('conv2d_5'),
conv6_params: extractConvParams('conv2d_6'),
conv7_params: extractConvParams('conv2d_7'),
fc0_params: extractFcParams('dense'),
fc1_params: extractFcParams('logits')
const params = {
conv0_params: extractConvParams('conv2d_0', 'conv0_params'),
conv1_params: extractConvParams('conv2d_1', 'conv1_params'),
conv2_params: extractConvParams('conv2d_2', 'conv2_params'),
conv3_params: extractConvParams('conv2d_3', 'conv3_params'),
conv4_params: extractConvParams('conv2d_4', 'conv4_params'),
conv5_params: extractConvParams('conv2d_5', 'conv5_params'),
conv6_params: extractConvParams('conv2d_6', 'conv6_params'),
conv7_params: extractConvParams('conv2d_7', 'conv7_params'),
fc0_params: extractFcParams('dense', 'fc0_params'),
fc1_params: extractFcParams('logits', 'fc1_params')
}
return { params, paramMappings }
}
\ No newline at end of file
test/tests/commons/NeuralNetwork.test.ts 0 → 100644
import { NeuralNetwork } from '../../../src/commons/NeuralNetwork';
import * as tf from '@tensorflow/tfjs-core';
class FakeNeuralNetwork extends NeuralNetwork<any> {
constructor(
convFilter: tf.Tensor = tf.tensor(0),
convBias: tf.Tensor = tf.tensor(0),
fcWeights: tf.Tensor = tf.tensor(0)
) {
super()
this._params = {
conv: {
filter: convFilter,
bias: convBias,
},
fc: fcWeights
}
this._paramMappings = [
{ originalPath: 'conv2d/filter', paramPath: 'conv/filter' },
{ originalPath: 'conv2d/bias', paramPath: 'conv/bias' },
{ originalPath: 'dense/weights', paramPath: 'fc' }
]
}
}
describe('NeuralNetwork', () => {
describe('getParamFromPath', () => {
it('returns correct params', () => tf.tidy(() => {
const convFilter = tf.tensor(0)
const convBias = tf.tensor(0)
const fcWeights = tf.tensor(0)
const net = new FakeNeuralNetwork(convFilter, convBias, fcWeights)
expect(net.getParamFromPath('conv/filter')).toEqual(convFilter)
expect(net.getParamFromPath('conv/bias')).toEqual(convBias)
expect(net.getParamFromPath('fc')).toEqual(fcWeights)
}))
it('throws if param is not a tensor', () => tf.tidy(() => {
const net = new FakeNeuralNetwork(null as any)
const fakePath = 'conv/filter'
expect(
() => net.getParamFromPath(fakePath)
).toThrowError(`traversePropertyPath - parameter is not a tensor, for path ${fakePath}`)
}))
it('throws if key path invalid', () => tf.tidy(() => {
const net = new FakeNeuralNetwork()
const fakePath = 'conv2d/foo'
expect(
() => net.getParamFromPath(fakePath)
).toThrowError(`traversePropertyPath - object does not have property conv2d, for path ${fakePath}`)
}))
})
describe('reassignParamFromPath', () => {
it('sets correct params', () => tf.tidy(() => {
const net = new FakeNeuralNetwork()
const convFilter = tf.tensor(0)
const convBias = tf.tensor(0)
const fcWeights = tf.tensor(0)
net.reassignParamFromPath('conv/filter', convFilter)
net.reassignParamFromPath('conv/bias', convBias)
net.reassignParamFromPath('fc', fcWeights)
expect(net.params.conv.filter).toEqual(convFilter)
expect(net.params.conv.bias).toEqual(convBias)
expect(net.params.fc).toEqual(fcWeights)
}))
it('throws if param is not a tensor', () => tf.tidy(() => {
const net = new FakeNeuralNetwork(null as any)
const fakePath = 'conv/filter'
expect(
() => net.reassignParamFromPath(fakePath, tf.tensor(0))
).toThrowError(`traversePropertyPath - parameter is not a tensor, for path ${fakePath}`)
}))
it('throws if key path invalid', () => tf.tidy(() => {
const net = new FakeNeuralNetwork()
const fakePath = 'conv2d/foo'
expect(
() => net.reassignParamFromPath(fakePath, tf.tensor(0))
).toThrowError(`traversePropertyPath - object does not have property conv2d, for path ${fakePath}`)
}))
})
describe('getParamList', () => {
it('returns param tensors with path', () => tf.tidy(() => {
const convFilter = tf.tensor(0)
const convBias = tf.tensor(0)
const fcWeights = tf.tensor(0)
const net = new FakeNeuralNetwork(convFilter, convBias, fcWeights)
const paramList = net.getParamList()
expect(paramList.length).toEqual(3)
expect(paramList[0].path).toEqual('conv/filter')
expect(paramList[1].path).toEqual('conv/bias')
expect(paramList[2].path).toEqual('fc')
expect(paramList[0].tensor).toEqual(convFilter)
expect(paramList[1].tensor).toEqual(convBias)
expect(paramList[2].tensor).toEqual(fcWeights)
}))
})
describe('getFrozenParams', () => {
it('returns all frozen params', () => tf.tidy(() => {
const convFilter = tf.tensor(0)
const convBias = tf.tensor(0)
const fcWeights = tf.variable(tf.scalar(0))
const net = new FakeNeuralNetwork(convFilter, convBias, fcWeights)
const frozenParams = net.getFrozenParams()
expect(frozenParams.length).toEqual(2)
expect(frozenParams[0].path).toEqual('conv/filter')
expect(frozenParams[1].path).toEqual('conv/bias')
expect(frozenParams[0].tensor).toEqual(convFilter)
expect(frozenParams[1].tensor).toEqual(convBias)
}))
})
describe('getTrainableParams', () => {
it('returns all trainable params', () => tf.tidy(() => {
const convFilter = tf.variable(tf.scalar(0))
const convBias = tf.variable(tf.scalar(0))
const fcWeights = tf.tensor(0)
const net = new FakeNeuralNetwork(convFilter, convBias, fcWeights)
const trainableParams = net.getTrainableParams()
expect(trainableParams.length).toEqual(2)
expect(trainableParams[0].path).toEqual('conv/filter')
expect(trainableParams[1].path).toEqual('conv/bias')
expect(trainableParams[0].tensor).toEqual(convFilter)
expect(trainableParams[1].tensor).toEqual(convBias)
}))
})
describe('dispose', () => {
it('disposes all param tensors', () => tf.tidy(() => {
const numTensors = tf.memory().numTensors
const net = new FakeNeuralNetwork()
net.dispose()
expect(net.params).toBe(undefined)
expect(tf.memory().numTensors - numTensors).toEqual(0)
}))
})
describe('variable', () => {
it('make all param tensors trainable', () => tf.tidy(() => {
const net = new FakeNeuralNetwork()
net.variable()
expect(net.params.conv.filter instanceof tf.Variable).toBe(true)
expect(net.params.conv.bias instanceof tf.Variable).toBe(true)
expect(net.params.fc instanceof tf.Variable).toBe(true)
}))
it('disposes old tensors', () => tf.tidy(() => {
const net = new FakeNeuralNetwork()
const numTensors = tf.memory().numTensors
net.variable()
expect(tf.memory().numTensors - numTensors).toEqual(0)
}))
})
describe('freeze', () => {
it('freezes all param variables', () => tf.tidy(() => {
const net = new FakeNeuralNetwork(
tf.variable(tf.scalar(0)),
tf.variable(tf.scalar(0)),
tf.variable(tf.scalar(0))
)
net.freeze()
expect(net.params.conv.filter instanceof tf.Variable).toBe(false)
expect(net.params.conv.bias instanceof tf.Variable).toBe(false)
expect(net.params.fc instanceof tf.Variable).toBe(false)
}))
it('disposes old tensors', () => tf.tidy(() => {
const net = new FakeNeuralNetwork(
tf.variable(tf.scalar(0)),
tf.variable(tf.scalar(0)),
tf.variable(tf.scalar(0))
)
const numTensors = tf.memory().numTensors
net.freeze()
expect(tf.memory().numTensors - numTensors).toEqual(0)
}))
})
})
tools/train/package.json 0 → 100644
{
"scripts": {
"start": "node server.js"
},
"author": "justadudewhohacks",
"license": "MIT",
"dependencies": {
"express": "^4.16.3",
"file-saver": "^1.3.8"
}
}
tools/train/public/faceLandmarksTrain.js 0 → 100644
async function promiseSequential(promises) {
const curr = promises[0]
if (!curr) {
return
}
await curr()
return promiseSequential(promises.slice(1))
}
async function trainStep(batchCreators) {
await promiseSequential(batchCreators.map((batchCreator, dataIdx) => async () => {
const { batchInput, landmarksBatchTensor } = await batchCreator()
let ts = Date.now()
const cost = optimizer.minimize(() => {
const out = window.trainNet.forwardInput(batchInput.managed())
const loss = lossFunction(
landmarksBatchTensor,
out
)
return loss
}, true)
ts = Date.now() - ts
console.log(`loss[${dataIdx}]: ${await cost.data()}, ${ts} ms (${ts / batchInput.batchSize} ms / batch element)`)
landmarksBatchTensor.dispose()
cost.dispose()
await tf.nextFrame()
}))
}
function createBatchCreators(data, batchSize) {
if (batchSize < 1) {
throw new Error('invalid batch size: ' + batchSize)
}
const batches = []
const pushToBatch = (remaining) => {
if (remaining.length) {
batches.push(remaining.slice(0, batchSize))
pushToBatch(remaining.slice(batchSize))
}
return batches
}
pushToBatch(data)
const batchCreators = batches.map(dataForBatch => async () => {
const imgs = dataForBatch.map(d => d.img)
const allLandmarks = dataForBatch.map(d => landmarkPositionsToArray(d.landmarks))
const batchInput = await faceapi.toNetInput(imgs)
const landmarksBatchTensor = tf.tidy(() => {
const landmarkTensors = allLandmarks.map(arr => tf.tensor2d(arr, [1, 136]))
return tf.stack(landmarkTensors, 0).as2D(-1, 136)
})
return {
batchInput,
landmarksBatchTensor
}
})
return batchCreators
}
function landmarkPositionsToArray(landmarks) {
return landmarks.getRelativePositions().map(pt => [pt.x, pt.y])
.reduce((flat, arr) => flat.concat(arr))
}
function toFaceLandmarks(landmarks, { naturalWidth, naturalHeight }) {
return new faceapi.FaceLandmarks(
landmarks.map(l => new faceapi.Point(l.x / naturalWidth, l.y / naturalHeight)),
{ width: naturalWidth, height: naturalHeight }
)
}
async function loadImagesInBatch(allLandmarks, offset = 0) {
return Promise.all(allLandmarks.map(async (landmarks, i) => {
const imgUri = `/train_images/${i + offset}.png`
const img = await faceapi.bufferToImage(await fetchImage(imgUri))
return {
imgUri,
img,
landmarks: toFaceLandmarks(landmarks, img)
}
}))
}
async function getTrainData() {
const landmarksJson = (await (await fetch('/train_landmarks.json')).json())
const numLandmarks = Object.keys(landmarksJson).length
const allLandmarks = Array.from(
new Array(numLandmarks),
(_, i) => landmarksJson[i]
)
const batch1 = await loadImagesInBatch(allLandmarks.slice(0, 4000))
const batch2 = await loadImagesInBatch(allLandmarks.slice(4000), 4000)
return batch1.concat(batch2)
}
// https://stackoverflow.com/questions/6274339/how-can-i-shuffle-an-array
function shuffle(a) {
var j, x, i;
for (i = a.length - 1; i > 0; i--) {
j = Math.floor(Math.random() * (i + 1));
x = a[i];
a[i] = a[j];
a[j] = x;
}
return a;
}
\ No newline at end of file
tools/train/public/faceLandmarksUi.js 0 → 100644
function onKeyDown(e) {
e.target.value = (
parseInt(e.target.value) + (e.keyCode === 38 ? 1 : (e.keyCode === 40 ? -1 : 0))
) || e.target.value || 0
const uri = '/train_images/' + e.target.value + '.png'
console.log(uri)
onSelectionChanged(uri)
}
function onChangeDrawLines(e) {
window.drawLines = $(e.target).prop('checked')
redraw()
}
async function onSelectionChanged(uri) {
const imgBuf = await fetchImage(uri)
window.currentImg = await faceapi.bufferToImage(imgBuf)
window.landmarks = await Promise.all(window.nets.map(
async net => net.detectLandmarks(window.currentImg)
))
redraw(uri)
}
function drawLandmarkCanvas(img, landmarks, drawOpts) {
const canvas = faceapi.createCanvasFromMedia(img)
$('#faceContainer').append(canvas)
faceapi.drawLandmarks(
canvas,
landmarks,
drawOpts
)
}
function redraw(uri) {
$('#faceContainer').empty()
const drawOpts = { lineWidth: window.drawLines ? 2 : 4, drawLines: window.drawLines }
window.landmarks.forEach(landmarks => {
drawLandmarkCanvas(window.currentImg, landmarks, drawOpts)
})
const trainImgAndLandmarks = window.trainData.find(t => t.imgUri === uri)
if (trainImgAndLandmarks) {
drawLandmarkCanvas(trainImgAndLandmarks.img, trainImgAndLandmarks.landmarks, drawOpts)
}
}
async function loadNet(file) {
const res = await fetch(file)
const weights = new Float32Array(await res.arrayBuffer())
return faceapi.faceLandmarkNet(weights)
}
async function init() {
//await faceapi.loadFaceLandmarkModel('/')
//window.nets.push(faceapi.landmarkNet)
//window.nets.push(await loadNet('retrained/landmarks_v0.weights'))
//window.nets.push(await loadNet('retrained/landmarks_v2.weights'))
window.trainNet = await loadNet('retrained/landmarks_v6.weights')
window.nets.push(trainNet)
$('#loader').hide()
await run()
await onSelectionChanged($('#selectList select').val())
}
$(document).ready(function() {
$('#imgByNr').keydown(onKeyDown);
renderFaceImageSelectList(
'#selectList',
onSelectionChanged,
{ className: 'sheldon', imageIdx: 1 }
)
init()
})
\ No newline at end of file
tools/train/server.js 0 → 100644
require('./.env')
const express = require('express')
const path = require('path')
const app = express()
const viewsDir = path.join(__dirname, 'views')
app.use(express.static(viewsDir))
app.use(express.static(path.join(__dirname, './public')))
app.use(express.static(path.join(__dirname, './tmp')))
app.use(express.static(path.join(__dirname, './node_modules/file-saver')))
app.use(express.static(path.join(__dirname, '../../examples/public')))
app.use(express.static(path.join(__dirname, '../../weights')))
app.use(express.static(path.join(__dirname, '../../dist')))
const trainDataPath = path.resolve(process.env.TRAIN_DATA_PATH)
app.use(express.static(trainDataPath))
app.get('/', (req, res) => res.redirect('/face_landmarks'))
app.get('/face_landmarks', (req, res) => res.sendFile(path.join(viewsDir, 'faceLandmarks.html')))
app.listen(3000, () => console.log('Listening on port 3000!'))
\ No newline at end of file
tools/train/views/faceLandmarks.html 0 → 100644
<!DOCTYPE html>
<html>
<head>
<link rel="stylesheet" href="styles.css">
<link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/materialize/0.100.2/css/materialize.css">
<script type="text/javascript" src="https://code.jquery.com/jquery-2.1.1.min.js"></script>
<script src="https://cdnjs.cloudflare.com/ajax/libs/materialize/0.100.2/js/materialize.min.js"></script>
<script src="face-api.js"></script>
<script src="commons.js"></script>
<script src="faceLandmarksUi.js"></script>
<script src="faceLandmarksTrain.js"></script>
<script src="FileSaver.js"></script>
</head>
<body>
<div id="navbar"></div>
<div class="center-content page-container">
<div>
<div class="progress" id="loader">
<div class="indeterminate"></div>
</div>
<div class="row side-by-side">
<div class="center-content">
<div id="faceContainer"></div>
<div id="selectList"></div>
</div>
</div>
<p>
<input type="checkbox" id="drawLinesCheckbox" checked="checked" onchange="onChangeDrawLines(event)" />
<label for="drawLinesCheckbox">Draw Lines</label>
</p>
<div class="row">
<label for="imgByNr">Enter image NR: </label>
<input id="imgByNr" type="text" class="bold">
</div>
</div>
</div>
<script>
tf = faceapi.tf
window.saveEveryNthIteration = 2
window.drawLines = true
window.nets = []
window.landmarks = []
window.trainSteps = 100
window.learningRate = 0.01
//window.optimizer = tf.train.sgd(learningRate)
window.optimizer = tf.train.adam(0.001, 0.9, 0.999, 1e-8)
function lossFunction(labels, out) {
return tf.losses.meanSquaredError(labels, out)
}
async function run() {
window.trainData = await getTrainData()
window.trainNet.variable()
return
await train()
}
async function train(batchSize = 10) {
for (let i = 0; i < trainSteps; i++) {
console.log('step', i)
const batchCreators = createBatchCreators(shuffle(window.trainData), batchSize)
let ts = Date.now()
await trainStep(batchCreators)
ts = Date.now() - ts
console.log('step %s done (%s ms)', i, ts)
if (((i + 1) % saveEveryNthIteration) === 0) {
saveWeights(i)
}
}
}
function saveWeights(idx = 0) {
const binaryWeights = new Float32Array(
window.trainNet.getParamList()
.map(({ tensor }) => Array.from(tensor.dataSync()))
.reduce((flat, arr) => flat.concat(arr))
)
saveAs(new Blob([binaryWeights]), 'landmark_trained_weights_' + idx + '.weights')
}
</script>
</body>
</html>
\ No newline at end of file
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