let faceapi;

let detections = [];

let storedFaces = []; // Stores recognized face landmarks

​

function setup() {

  createCanvas(640, 480);

  video = createCapture(VIDEO);

  video.size(width, height);

  video.hide();

​

  // Load FaceAPI models

  faceapi = ml5.faceApi(video, { withLandmarks: true, withDescriptors: true }, modelLoaded);

}

​

function modelLoaded() {

  console.log("FaceAPI model loaded!");

  detectFace();

}

​

function detectFace() {

  faceapi.detect(gotResults);

}

​

function gotResults(err, result) {

  if (err) {

    console.error(err);

    return;

  }

​

  detections = result;

  if (detections.length > 0) {

    let descriptor = detections[0].descriptor; // Unique 128D face vector

​

    let match = findMatchingFace(descriptor);

    if (match === null) {

      storedFaces.push(descriptor);

      console.log("New face stored! Total stored:", storedFaces.length);

    } else {

      console.log("Recognized a stored face!");

    }

  }

  // remove old faces to clear memory

  if (storedFaces.length > 10) {

    for (let i; i<9; i++){

      storedFaces.shift();

    }

  }

​

  detectFace(); // Loop face detection

}

​

function findMatchingFace(newDescriptor) {

  for (let stored of storedFaces) {

    let similarity = cosineSimilarity(stored, newDescriptor);

    if (similarity > 0.9) { // Threshold for recognition

      return stored;

    }

  }

  return null;

}

​

function cosineSimilarity(vec1, vec2) {

  let dot = 0, mag1 = 0, mag2 = 0;

  for (let i = 0; i < vec1.length; i++) {

    dot += vec1[i] * vec2[i];

    mag1 += vec1[i] ** 2;

    mag2 += vec2[i] ** 2;

  }

  return dot / (Math.sqrt(mag1) * Math.sqrt(mag2));

}

​

function draw() {

  image(video, 0, 0, width, height);

  if (detections.length > 0) {

    let points = detections[0].landmarks.positions;

    for (let p of points) {

      fill(0, 255, 0);

      noStroke();

      ellipse(p._x, p._y, 5, 5);

    }

  }

}

​