API notes

Require files

• lib/face-api.min.js
• Related moodels in models/*
• src/face_status.js

Using example

Import files js in index.html

  <script defer src="lib/face-api.min.js"></script>
  <script defer type="module" src="src/script.js"></script>

Run script in in src/script.js

Load models

Promise.all([
    faceapi.nets.ssdMobilenetv1.loadFromUri('/models'),
    faceapi.nets.faceLandmark68Net.loadFromUri('/models'),
    faceapi.nets.faceRecognitionNet.loadFromUri('/models'),
    faceapi.nets.faceExpressionNet.loadFromUri('/models'),
]).then(startVideo)

Start video

function startVideo(){
    navigator.mediaDevices.getUserMedia({ video: true })
    .then(stream => video.srcObject = stream)
    .then(() => new Promise(resolve => video.onloadedmetadata = resolve))
  
}

Run face status recognize

video.addEventListener('play',() => {
    let canvas,displaySize;
    [canvas,displaySize] = face_status.init(video);
    const oval = {
        x : 300,
        y : 100,
        w : 180,
        h : 200,
    }
    result = face_status.detect_face_status(video,canvas,displaySize,oval);
}

Return result

result is a string in following format:

face_status1,face_status2,face_status3

Example

happy,turn_left,in_oval