// A Beginner's Guide to Machine Learning with ml5.js

// The Coding Train / Daniel Shiffman

// https://youtu.be/eeO-rWYFuG0

// https://thecodingtrain.com/learning/ml5/3.1-feature-extractor-classification.html

// https://editor.p5js.org/codingtrain/sketches/5A_TJHA1

​

// (This sketch is not working in safari)

​

let mobilenet;

let classifier;

let video;

let label = 'test';

let ukeButton;

let whistleButton;

let trainButton;

​

function modelReady() {

  console.log('Model is ready!!!');

}

​

function videoReady() {

  console.log('Video is ready!!!');

}

​

function whileTraining(loss) {

  if (loss == null) {

    console.log('Training Complete');

    classifier.classify(gotResults);

  } else {

    console.log(loss);

  }

}

​

function gotResults(error, result) {

  if (error) {

    console.error(error);

  } else {

    // updated to work with newer version of ml5

    // label = result;

    label = result[0].label;

    classifier.classify(gotResults);

  }

}

​

function setup() {

  createCanvas(320, 270);

  video = createCapture(VIDEO);

  video.hide();

  background(0);

  mobilenet = ml5.featureExtractor('MobileNet', modelReady);

  classifier = mobilenet.classification(video, videoReady);

​

  ukeButton = createButton('happy');

  ukeButton.mousePressed(function() {

    classifier.addImage('happy');

  });

​

  whistleButton = createButton('sad');

  whistleButton.mousePressed(function() {

    classifier.addImage('sad');

  });

​

  trainButton = createButton('train');

  trainButton.mousePressed(function() {

    classifier.train(whileTraining);

  });

}

​

function draw() {

  background(0);

  image(video, 0, 0, 320, 240);

  fill(255);

  textSize(16);

  text(label, 10, height - 10);

}

​