let submitButton;

​

let rSlider, gSlider, bSlider;

let labelP;

let lossP;

​

function setup() {

  // Crude interface

  lossP = createP('loss');

​

  createCanvas(100, 100);

​

  labelP = createP('label');

​

  rSlider = createSlider(0, 255, 255);

  gSlider = createSlider(0, 255, 0);

  bSlider = createSlider(0, 255, 255);

​

  let nnOptions = {

    task: 'classification',

    debug: true

  };

  neuralNetwork = ml5.neuralNetwork(nnOptions);

​

  // option 1: Load model explictly pointing to each file

  const modelDetails = {

    model: 'model/model.json',

    metadata: 'model/model_meta.json',

    weights: 'model/model.weights.bin'

  }

  neuralNetwork.load(modelDetails, modelReady);

  // option 2: Load model just pointing to the model file

  // neuralNetwork.load('model/model.json', modelReady);

}

​

function modelReady() {

  console.log('model loaded!')

  classify();

};

​

​

function classify() {

  let inputs = {

    r: rSlider.value(),

    g: gSlider.value(),

    b: bSlider.value()

  }

  neuralNetwork.classify(inputs, gotResults);

}

​

function gotResults(error, results) {

  if (error) {

    console.error(error);

  } else {

    // console.log(results)

    labelP.html(`label:${results[0].label}, confidence: ${results[0].confidence.toFixed(2)}`);

    classify();

  }

}

​

function draw() {

  background(rSlider.value(), gSlider.value(), bSlider.value());

}