// www.ernst-schmidt.com

​

// idea: https://www.youtube.com/watch?v=KO7W0Qq8yUE&feature=youtu.be

// most relevant source: https://www.youtube.com/playlist?list=PLRqwX-V7Uu6Y7MdSCaIfsxc561QI0U0Tb

​

// open file tree on the left to see all files!

​

let r, g, b;

​

function setup() {

  createCanvas(400, 400);

  yBorderBottom = height - 100;

  brain = new NeuralNetwork(3, 3, 2);

  brainDisplay = new NNVisualization(brain, 60, yBorderBottom + 20);

  nTrainings = 0;

  trainingLog = [];

  guess = [];

  accuracy = undefined;

​

  randomClr();

  predict(r, g, b);

  evaluate();

}

​

function draw() {

  displayWindow();

  displayWheel();

  displayInfobox();

  displayNTrainings();

  displayGuess();

  displayAccuracy();

  displayClr();

  brainDisplay.display();

}

​

function predict(r, g, b) {

  let inputs = [r / 255, g / 255, b / 255];

  guess = brain.guess(inputs);

}

​

function randomClr() {

  if (random() < .5) {

    r = floor(random(0, 50));

    g = floor(random(0, 50));

    b = floor(random(0, 50));

  } else {

    r = floor(random(201, 256));

    g = floor(random(201, 256));

    b = floor(random(201, 256));

  }

}

​

function evaluate() {

  let sumRGB = r + g + b;

  let result = false;

  if (sumRGB < 765 / 2 && guess[0] > guess[1]) {

    result = true;

  } else if (sumRGB < 765 / 2 && guess[0] < guess[1]) {

    result = false;

  } else if (sumRGB > 765 / 2 && guess[0] < guess[1]) {

    result = true;

  } else if (sumRGB > 765 / 2 && guess[0] > guess[1]) {

    result = false;

  }

​

  if (result) {

    trainingLog.push(1);

  } else {

    trainingLog.push(0);

  }

​

  if (trainingLog.length > 20) {

    trainingLog.splice(0, 1);

  }

​

  let sumResults = 0;

  if (trainingLog.length >= 20) {

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

      sumResults += trainingLog[i];

    }

    accuracy = sumResults / 20;

  }

}