let targetLabel='1'

​

let state='collection'

​

function setup() {

  createCanvas(400, 400);

  let options ={

  inputs: ["x","y"],

  outputs: ['label'],

  task: 'classification',

 // debug: 'true'

  }

  NNmodel=ml5.neuralNetwork(options)

  background(245)

}

​

​

function keyPressed(){

targetLabel=key.toUpperCase()

  if(key=='t'){

    state='training'

    console.log("Starting Training...")

    NNmodel.normalizeData()

   let options={

    epochs: 1000

    }

  NNmodel.train(options, whileTraining, finishedTraining)

  }

}

​

function mousePressed(){

  let inputs={

  x:mouseX,

  y:mouseY

  }

  if (state=='collection'){ 

  let target={

  label:targetLabel

  }

NNmodel.addData(inputs,target)

stroke(255,0,0)

fill(255,0,0,100)

ellipse(mouseX,mouseY,20)

fill(0)

noStroke()

textAlign(CENTER,CENTER)

text(targetLabel,mouseX,mouseY)

}else if(state=='prediction'){

NNmodel.classify(inputs,gotResults)

}

}

function whileTraining(epoch, loss){

console.log("Training..","Epoch:",epoch,"Loss:",loss)

  background(245)

textSize(20)

text(round((epoch+1)/10),180,200)

  text("/100",220,200)

}

​

function finishedTraining(){

  background(245)

  state='prediction'

    console.log('Finished Training')

​

}

​

function gotResults(error,results){

if (error){

console.error(error)

  return;

}

  console.log(results)

  stroke(0,200,255)

fill(0,200,255,100)

ellipse(mouseX,mouseY,20)

fill(0)

noStroke()

textAlign(CENTER,CENTER)

text(results[0].label,mouseX,mouseY)

}

​