// NYUSH F24 - gohai

​

let curColor = "green";

let nn;

let training = true;

let classifyX;

let classifyY;

let resultColor = "white";

let resultX = 0;

let resultY = 0;

​

function setup() {

  createCanvas(400, 400);

​

  // For this example to work across all browsers

  // "webgl" or "cpu" needs to be set as the backend

  ml5.setBackend("webgl");

​

  let options = {

    // object literal

    task: "classification",

    debug: true,

  };

​

  nn = ml5.neuralNetwork(options);

}

​

function draw() {

  background(255);

​

  let points = nn.neuralNetworkData.data.raw;

​

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

    noStroke();

    fill(points[i].ys[0]);

    ellipse(points[i].xs[0], points[i].xs[1], 8, 8);

  }

​

  if (training == false) {

    fill(resultColor);

    circle(resultX, resultY, 16);

  }

}

​

function mousePressed() {

  if (training) {

    // loop to synthesize a couple of

    // samples around the mouse position

    for (let i = 0; i < 8; i++) {

      let inputs = [mouseX + random(-8, 8), mouseY + random(-8, 8)];

      let outputs = [curColor];

      nn.addData(inputs, outputs);

    }

  } else {

    let inputs = [mouseX, mouseY];

    nn.classify(inputs, doneClassifying);

    classifyX = mouseX;

    classifyY = mouseY;

  }

}

​

function keyPressed() {

  if (key == "g" || key == "G") {

    curColor = "green";

  } else if (key == "r" || key == "R") {

    curColor = "red";

  } else if (key == "b" || key == "B") {

    curColor = "blue";

  } else if (key == "s") {

    nn.save(); // downloads the trained model

  } else if (key == "S") {

    nn.saveData(); //  downloads the training data

  } else if (key == "l") {

    // loads the model

    nn.load(

      {

        model: "model.json",

        metadata: "model_meta.json",

        weights: "model.weights.bin",

      },

      finishedLoading

    );

  } else if (key == "L") {

    nn.loadData("data.json"); // loads the training data

  } else if (key == "t") {

    nn.normalizeData();

​

    let options = {

      epochs: 64,

      batchSize: 12,

    };

​

    nn.train(options, doneTraining);

  }

}

​

function doneTraining() {

  console.log("Done training!");

  training = false;

}

​

function doneClassifying(results) {

  //console.log(results);

  resultColor = results[0].label;

  resultX = classifyX;

  resultY = classifyY;

}

​

function finishedLoading() {

  console.log("Loaded model!");

  training = false;

}

​