// Cloned by Jack O'Brien on 5 Dec 2020 from World "Character recognition neural network" by "Coding Train" project
// Please leave this clone trail here.
// Port of Character recognition neural network from here:
// https://github.com/CodingTrain/Toy-Neural-Network-JS/tree/master/examples/mnist
// with many modifications
// --- defined by MNIST - do not change these ---------------------------------------
const PIXELS = 28; // images in data set are tiny
const PIXELSSQUARED = PIXELS * PIXELS;
// number of training and test exemplars in the data set:
const NOTRAIN = 60000;
const NOTEST = 10000;
//--- can modify all these --------------------------------------------------
// no of nodes in network
const noinput = PIXELSSQUARED;
const nohidden1 = 64;
const nohidden2 = 64;
const nooutput = 10;
const learningrate = 0.01; // default 0.1
// should we train every timestep or not
let do_training = true;
// how many to train and test per timestep
const TRAINPERSTEP = 32;
const TESTPERSTEP = 16;
// multiply it by this to magnify for display
const ZOOMFACTOR = 10;
const ZOOMPIXELS = ZOOMFACTOR * PIXELS;
// 3 rows of
// large image + 50 gap + small image
// 50 gap between rows
const canvaswidth = ( PIXELS + ZOOMPIXELS ) + 415;
const canvasheight = ( ZOOMPIXELS * 3 ) + 100;
const DOODLE_THICK = 16; // thickness of doodle lines
const DOODLE_BLUR = 3; // blur factor applied to doodles
let mnist;
// all data is loaded into this
// mnist.train_images
// mnist.train_labels
// mnist.test_images
// mnist.test_labels
let nn;
let trainrun = 1;
let train_index = 0;
let testrun = 1;
let test_index = 0;
let total_tests = 0;
let total_correct = 0;
let globalScore = 0;
let learningRateFlag = false;
// images in LHS:
let doodle, demo;
let doodle_exists = false;
let demo_exists = false;
let mousedrag = false; // are we in the middle of a mouse drag drawing?
// ConfusionMatrix
let confusionM;
let canvas;
// let draw;
let ctxdraw;
var thedoodleimage;
var doodleImageData;
var convnet;
let trainer;
const inputSize = 24;
// save inputs to global var to inspect
// type these names in console
var train_inputs, test_inputs, demo_inputs, doodle_inputs;
// Matrix.randomize() is changed to point to this. Must be defined by user of Matrix.
function randomWeight()
{
return ( AB.randomFloatAtoB ( -0.9, 0.9 ) );
// Coding Train default is -1 to 1
}
// CSS trick
// make run header bigger
$("#runheaderbox").css ( { "max-height": "95vh" } );
//--- start of AB.msgs structure: ---------------------------------------------------------
// We output a serious of AB.msgs to put data at various places in the run header
var thehtml;
// 1 Doodle header
thehtml = "<hr> <h1> 1. Doodle </h1> Top row: Doodle (left) and shrunk (right). <br> " +
" Draw your doodle in top LHS. <button onclick='wipeDoodle();' class='normbutton' >Clear doodle</button> <br> " +
"<div></div>";
// <canvas id='doodleCanvas' style='width:358px;height:400px'></canvas>
AB.msg ( thehtml, 1 );
// 2 Doodle variable data (guess)
// 3 Training header
thehtml = "<hr> <h1> 2. Training </h1> Middle row: Training image magnified (left) and original (right). <br> " +
" <button onclick='do_training == false?do_training = true: do_training = false;' class='normbutton' >Toggle training</button> " +
" <button class='normbutton' onclick='saveModel()'>Save Model</button><button class='normbutton' onclick='loadModel()'>Load Model</button> <br> ";
AB.msg ( thehtml, 3 );
// 4 variable training data
// 5 Testing header
thehtml = "<h3> Hidden tests </h3> " ;
AB.msg ( thehtml, 5 );
// 6 variable testing data
// 7 Demo header
thehtml = "<hr> <h1> 3. Demo </h1> Bottom row: Test image magnified (left) and original (right). <br>" +
" The network is <i>not</i> trained on any of these images. <br> " +
" <button onclick='makeDemo();' class='normbutton' >Demo test image</button> <br>";
AB.msg ( thehtml, 7 );
// 8 Demo variable data (random demo ID)
// 9 Demo variable data (changing guess)
const greenspan = "<span style='font-weight:bold; font-size:x-large; color:darkgreen'> " ;
//--- end of AB.msgs structure: ---------------------------------------------------------
function setup()
{
createCanvas ( canvaswidth, canvasheight );
doodle = createGraphics ( ZOOMPIXELS, ZOOMPIXELS ); // doodle on larger canvas
doodle.pixelDensity(1);
// draw = document.getElementById("defaultCanvas0");
// ctxdraw = draw.getContext("2d");
// painting = false,
// lastX = 0,
// lastY = 0,
// lineThickness = 1;
// JS load other JS
// maybe have a loading screen while loading the JS and the data set
AB.loadingScreen();
$.getScript ( "/uploads/jobrien14/webcnn.js", function()
{
$.getScript ( "/uploads/jobrien14/mathutils.js", function()
{
$.getScript ( "/uploads/jobrien14/mnist.js", function()
{
$.getScript ( "/uploads/jobrien14/matrix.js", function()
{
console.log ("All JS loaded");
createModel();
confusionM = new Matrix(10, 10);
loadData();
});
});
});
});
}
//Code to create initial model architecture
function createModel(){
convnet = new WebCNN;
convnet.newLayer( { name: "image", type: LAYER_TYPE_INPUT_IMAGE, width: inputSize, height: inputSize, depth: 1 } );
convnet.newLayer( { name: "conv1", type: LAYER_TYPE_CONV, units: 10, kernelWidth: 5, kernelHeight: 5, strideX: 1, strideY: 1, padding: false, activation: ACTIVATION_RELU } );
convnet.newLayer( { name: "pool1", type: LAYER_TYPE_MAX_POOL, poolWidth: 2, poolHeight: 2, strideX: 2, strideY: 2 } );
convnet.newLayer( { name: "conv2", type: LAYER_TYPE_CONV, units: 20, kernelWidth: 5, kernelHeight: 5, strideX: 1, strideY: 1, padding: false, activation: ACTIVATION_RELU } );
convnet.newLayer( { name: "pool2", type: LAYER_TYPE_MAX_POOL, poolWidth: 2, poolHeight: 2, strideX: 2, strideY: 2 } );
convnet.newLayer( { name: "out", type: LAYER_TYPE_FULLY_CONNECTED, units: 10, activation: ACTIVATION_SOFTMAX } );
convnet.initialize();
convnet.setLearningRate( 0.01 );
convnet.setMomentum( 0.9 );
convnet.setLambda( 0.0 );
return convnet;
}
function saveModel(){
console.log('Saving Model to server')
AB.saveData ( nn.serialize());
}
function loadModel()
{
AB.restoreData ( function ( a )
{
console.log('Restoring')
// nn.deserialize(a);
});
}
// load data set from local file (on this server)
function loadData()
{
loadMNIST ( function(data)
{
mnist = data;
console.log ("All data loaded into mnist object:")
console.log(mnist);
AB.removeLoading(); // if no loading screen exists, this does nothing
});
}
function getImage ( img ) // make a P5 image object from a raw data array
{
let theimage = createImage (PIXELS, PIXELS); // make blank image, then populate it
theimage.loadPixels();
for (let i = 0; i < PIXELSSQUARED ; i++)
{
let bright = img[i];
let index = i * 4;
theimage.pixels[index + 0] = bright;
theimage.pixels[index + 1] = bright;
theimage.pixels[index + 2] = bright;
theimage.pixels[index + 3] = 255;
}
theimage.updatePixels();
return theimage;
}
function getInputs ( img ) // convert img array into normalised input array
{
let inputs = [];
for (let i = 0; i < PIXELSSQUARED ; i++)
{
let bright = img[i];
inputs[i] = bright / 255; // normalise to 0 to 1
}
return ( inputs );
}
// Image processing
// returns the image in a suitable format that the model expects
// expects a 3 dimensional object with a height, width and pixel data, must also match model input
function imagePreProcessing(img, Dimensions) {
return {
width: Dimensions,
height: Dimensions,
data: getImage(randomCrop(img, Dimensions), Dimensions).pixels
}
}
// Randomly crops the input image so as to reduce overfitting and imporve generalisation
// Mentioned on the associate website and github
function randomCrop(img, Dimensions) {
const cropVal = PIXELS - Dimensions;
var randVal1 = Math.floor(Math.random() * cropVal);
var randVal2 = Math.floor(Math.random() * cropVal);
let newWidth = randVal1 + Dimensions;
let newHeight = randVal2 + Dimensions;
let outputImg = [];
for (let i = randVal1; i < newWidth; i++)
for (let j = randVal2; j < newHeight; j++)
outputImg.push(img[i * PIXELS + j]);
return outputImg
}
function trainit (show) // train the network with a single exemplar, from global var "train_index", show visual on or off
{
if (train_index % TRAINPERSTEP !== 0)
return void train_index++;
let img = mnist.train_images[train_index];
let label = mnist.train_labels[train_index];
let trainImages = [];
let trainLabels = [];
// For this implementation the model is trained in batches as opposed to on single images at a time like the original implementation
// Starting from the current train_index, a batch of images are loaded into an array as this is what the model expects as an input
for(var i =0;i< TRAINPERSTEP;i++){
trainImages.push(imagePreProcessing(mnist.train_images[train_index+i], inputSize));
trainLabels.push(mnist.train_labels[train_index+i]);
}
var theimage = getImage ( img ); // get image from data array
image ( theimage, 0, ZOOMPIXELS+50, ZOOMPIXELS, ZOOMPIXELS ); // magnified
image ( theimage, ZOOMPIXELS+50, ZOOMPIXELS+50, PIXELS, PIXELS ); // original
// The model is trained on the current batch with the given labels
convnet.trainCNNClassifier(trainImages, trainLabels);
thehtml = " trainrun: " + trainrun + "<br> no: " + train_index ;
AB.msg ( thehtml, 4 );
train_index++;
if ( train_index == NOTRAIN )
{
train_index = 0;
console.log( "finished trainrun: " + trainrun );
trainrun++;
}
}
let img1;
function testit() // test the network with a single exemplar, from global var "test_index"
{
let img = mnist.test_images[test_index];
let label = mnist.test_labels[test_index];
img1=img;
// set up the inputs
let inputs = getInputs ( img );
test_inputs = inputs; // can inspect in console
// similar to above, the image needs to be put through the same process as the training data to have the best possibility for a match
let prediction = convnet.classifyImages([imagePreProcessing(img, inputSize)]); // array of outputs
let guess = findMax(prediction); // the top output
// let confidence = 100;
//
confusionM.data[guess][label]++;
total_tests++;
if (guess == label) total_correct++;
let percent = (total_correct / total_tests) * 100 ;
globalScore = percent
thehtml = " testrun: " + testrun + "<br> no: " + total_tests + " <br> " +
" correct: " + total_correct + "<br>" +
" score: " + greenspan + percent.toFixed(2) + "</span>";
AB.msg ( thehtml, 6 );
test_index++;
if ( test_index == NOTEST )
{
console.log( "finished testrun: " + testrun + " score: " + percent.toFixed(2) );
confusionM.print();
confusionM = new Matrix(10, 10);
testrun++;
test_index = 0;
total_tests = 0;
total_correct = 0;
}
}
//--- find no.1 (and maybe no.2) output nodes ---------------------------------------
// (restriction) assumes array values start at 0 (which is true for output nodes)
let no1value;
let no2value;
let no1;
let no2;
// function was editied to work with the convnet model and to find the top 2 predictions
function find12 (a) // return array showing indexes of no.1 and no.2 values in array
{
no1 = 0;
no2 = 0;
// Find no1
for(var i=0;i<a[0].size;i++){
a[0].getValue(0, 0, i)>a[0].getValue(0, 0, no1)?no1=i:no1=no1;
}
// Find no2
for(var j=0;j<a[0].size;j++){
(a[0].getValue(0, 0, j)>a[0].getValue(0, 0, no2) && a[0].getValue(0, 0, j)<a[0].getValue(0, 0, no1))?no2=j:no2=no2;
}
no1value = a[0].getValue(0, 0, no1);
no2value = a[0].getValue(0, 0, no2);
var b = [[ no1, no1value ], [no2, no2value]];
return b;
}
// just get the maximum - separate function for speed - done many times
// find our guess - the max of the output nodes array
// Altered similar to above, uses built in .get Value to find the score for each of the nodes and returns the highest score
function findMax (a)
{
let test= a[0];
var guess =0;
for(var i=0;i<test.size;i++){
test.getValue(0, 0, i)>test.getValue(0, 0, guess)?guess=i:guess=guess
}
return guess;
}
// --- the draw function -------------------------------------------------------------
// every step:
function draw()
{
// check if libraries and data loaded yet:
if ( typeof mnist == 'undefined' ) return;
// how can we get white doodle on black background on yellow canvas?
// background('#ffffcc'); doodle.background('black');
// background ('black');
stroke(0);
fill(0);
rect(0, 0, ( PIXELS + ZOOMPIXELS ) + 50, ( ZOOMPIXELS * 3 ) + 100);
fill(255);
rect(( PIXELS + ZOOMPIXELS ) + 58, ( PIXELS + ZOOMPIXELS ) + 58, ( PIXELS + ZOOMPIXELS ) + 50, ( ZOOMPIXELS * 3 ) + 100);
if ( do_training )
{
for (let i = 0; i < TRAINPERSTEP; i++)
{
if (i == 0){
trainit(true);
} // show only one per step - still flashes by
else
trainit(false);
}
// do some testing per step
for (let i = 0; i < TESTPERSTEP; i++)
testit();
}
// keep drawing demo and doodle images
// and keep guessing - we will update our guess as time goes on
if ( demo_exists )
{
drawDemo();
guessDemo();
}
if ( doodle_exists )
{
drawDoodle();
guessDoodle();
}
// detect doodle drawing
// (restriction) the following assumes doodle starts at 0,0
if ( mouseIsPressed ) // gets called when we click buttons, as well as if in doodle corner
{
// console.log ( mouseX + " " + mouseY + " " + pmouseX + " " + pmouseY );
var MAX = ZOOMPIXELS + 20; // can draw up to this pixels in corner
if ( (mouseX < MAX) && (mouseY < MAX) && (pmouseX < MAX) && (pmouseY < MAX) )
{
mousedrag = true; // start a mouse drag
doodle_exists = true;
doodle.stroke('white');
doodle.strokeWeight( DOODLE_THICK );
doodle.line(mouseX, mouseY, pmouseX, pmouseY);
}
}
else
{
// are we exiting a drawing
if ( mousedrag )
{
mousedrag = false;
// console.log ("Exiting draw. Now blurring.");
doodle.filter (BLUR, DOODLE_BLUR); // just blur once
// doodle.filter (ERODE);
// doodle.filter (ERODE);
// doodle.filter (ERODE);
// doodle.filter (ERODE);
// doodle.filter (ERODE);
console.log (doodle);
}
}
textSize(22);
fill(255);
text("Confusion Matrix", ( PIXELS + ZOOMPIXELS ) + 140, ( PIXELS + ZOOMPIXELS)+50);
for(var i = 0;i<10;i++){
for(var j = 0;j<10;j++){
var innerValue = confusionM.data[i][j];
var x = (( PIXELS + ZOOMPIXELS ) + 58)+(i*35);
var y = (( PIXELS + ZOOMPIXELS ) + 58)+(j*35);
// stroke(0);
squareColor = color(80, 20, 255);
squareColor.setAlpha(255*(innerValue/1000));
fill(squareColor);
rect(x, y, 35, 35);
textSize(12);
fill(0);
text(innerValue, x+10, y+25);
}
}
}
//--- demo -------------------------------------------------------------
// demo some test image and predict it
// get it from test set so have not used it in training
function makeDemo()
{
demo_exists = true;
var i = AB.randomIntAtoB ( 0, NOTEST - 1 );
demo = mnist.test_images[i];
var label = mnist.test_labels[i];
thehtml = "Test image no: " + i + "<br>" +
"Classification: " + label + "<br>" ;
AB.msg ( thehtml, 8 );
// type "demo" in console to see raw data
}
function drawDemo()
{
var theimage = getImage ( demo );
// console.log (theimage);
image ( theimage, 0, canvasheight - ZOOMPIXELS, ZOOMPIXELS, ZOOMPIXELS ); // magnified
image ( theimage, ZOOMPIXELS+50, canvasheight - ZOOMPIXELS, PIXELS, PIXELS ); // original
}
function guessDemo()
{
let inputs = getInputs ( demo );
demo_inputs = inputs; // can inspect in console
let prediction = convnet.classifyImages([imagePreProcessing(demo, inputSize)]);
let guess = findMax(prediction);
thehtml = " We classify it as: " + greenspan + guess + "</span>" ;
AB.msg ( thehtml, 9 );
}
let doodle1;
//--- doodle -------------------------------------------------------------
function drawDoodle()
{
// doodle is createGraphics not createImage
let theimage = doodle.get();
// console.log (theimage);
image ( theimage, 0, 0, ZOOMPIXELS, ZOOMPIXELS ); // original
image ( theimage, ZOOMPIXELS+50, 0, PIXELS, PIXELS ); // shrunk
}
function formatDoodle(img, Dimensions){
let outputImg = [];
for (let i = 0; i < Dimensions; i++)
for (let j = 0; j < Dimensions; j++)
outputImg.push(img[i * PIXELS + j]);
// return outputImg
return {
width: Dimensions,
height: Dimensions,
data: getImage(outputImg).pixels
}
}
function guessDoodle()
{
if(mousedrag===false)return;
// doodle is createGraphics not createImage
let img = doodle.get();
img.resize ( PIXELS, PIXELS );
img.loadPixels();
doodle1=img;
let inputs = [];
for (let j = 0; j < PIXELS; j++) {
/** @type {!Array} */
inputs[j] = [];
for (let i = 0; i < PIXELS; i++) {
inputs[j][i] = img.pixels[4 * (j * PIXELS + i)];
}
}
// doodle_inputs = inputs; // can inspect in console
// thedoodleimage = getImage ( doodle_inputs );
// doodleImageData = thedoodleimage.imageData;
// feed forward to make prediction
let prediction = convnet.classifyImages([formatDoodle(inputs, inputSize)]);
let b = find12(prediction);
thehtml = " We classify it as: " + greenspan + b[0][0] + "</span> with a confidence of: " + (b[0][1]*100).toFixed(2) + "%<br>" +
" No.2 guess is: " + greenspan + b[1][0] + "</span> with a confidence of: " + (b[1][1]*100).toFixed(2) + "%";
AB.msg ( thehtml, 2 );
}
function wipeDoodle()
{
doodle_exists = false;
doodle.background('black');
}
// --- debugging --------------------------------------------------
// in console
// showInputs(demo_inputs);
// showInputs(doodle_inputs);
function showInputs ( inputs )
// display inputs row by row, corresponding to square of pixels
{
var str = "";
for (let i = 0; i < inputs.length; i++)
{
if ( i % PIXELS == 0 ) str = str + "\n"; // new line for each row of pixels
var value = inputs[i];
str = str + " " + value.toFixed(2) ;
}
console.log (str);
}
// Penicl Drawing