Some fixes

fixed memory leaks in extractFaces, imageTensorToCanvas, toNetInput with batch size 1 tensor4d array + removed tests using createFakeHTMLVideoElement as it seems to not load the example video properly some times

added general handling of batch inputs + face landmark net accepts batch inputs now

added some tests to indentify memory leaks in NetInput and face landmark net + fixed leak in detectLandmarks

added batch input example for face landmarks + fixed: return array if batch input with array of length 1

test cases for batch input creation + e2e face landmark net batch input tests + added some face landmark net tests for non squared input

pad face landmark net input to square and center to prevent stretching of input image

Join Slack Community: Briim - Machine Learning in JavaScript

padToSquare fix

make padToSquare padding order append first when size is uneven + fixed issue when dimensions differ by 1 only

Sourcemaps

fix typo in README

wait for model being loaded before processing video frames

Await media loaded

use async toNetInput instead creating NetInput directly, which awaits all media elements of the input to be finished loading

faceapi.allFaces + global api