Skip to content

A project on making a game like Quick Draw using Machine Learning with Convolutional Neural Network.

License

Notifications You must be signed in to change notification settings

ssusnic/Machine-Learning-Doodle-Recognition

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

40 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Machine Learning: How to make Quick Draw game using Convolutional Neural Network

A Javascript Machine Learning Project on making a game like Quick, Draw!. It is an online game that challenges players to draw a doodle and then artificial intelligence guesses what the drawings represent.

This repo contains source code for implementing this game step by step.

Here is the video in which I explained the entire project:
https://youtu.be/AsYczuDBGp0

Project Info

The game is made in HTML5 using Phaser 2 framework and TensorFlow.js library for machine learning.

To classify images, I implemented an Artificial Intelligence (AI) based on Machine Learning (ML) and Convolutional Neural Network (CNN).

The model was trained on a small subset of the Quick Draw Dataset.

Here is a screenshot of the fully completed project:

Doodle Predictor Screenshot

To play the game online, visit my official website:
https://www.askforgametask.com/tutorial/machine-learning/quick-draw/

To read more tutorials and play my other games, check the homepage:
https://www.askforgametask.com

Running the Game

To play this game locally, you need to run it in a browser using a local web server as follows:

  1. install, configure and run a web server: for instance, let's suppose your server is XAMPP installed in C:\Xampp

  2. navigate to the server document root: C:\Xampp\htdocs

  3. create a new folder called 'doodle_predictor': C:\Xampp\htdocs\doodle_predictor

  4. download the project

  5. copy all project files directly in C:\Xampp\htdocs\doodle_predictor

  6. now you should have the following folder structure under C:\Xampp\htdocs\doodle_predictor:
    \assets (with game resources)
    \data (with datasets)
    \libs (with libraries)
    \part1 (with source files of the part 1)
    ...

  7. open a web browser and navigate to the http://localhost:<port>/doodle_predictor/part1
    to run the game of the part 1

Parts of the Project

This project is divided into 8 parts. Each part contains a fully commented source code so you can easily follow the game development step by step.

Here are short descriptions of each part:

Part 1: Setup

  • creating the Main Program with the main state
  • defining all substates of the main state
  • building the basic game structure with the main loop where the entire game logic will be implemented
  • creating a new User Interface class to allow users an interaction with the game

Part 2: Getting Data

  • creating a new CNN class for implementing a Convolutional Neural Network model
  • loading three datasets (car, fish and snowman)
  • splitting datasets into training and test data
  • shuffling data

Part 3: Building the Model

  • creating a sequential CNN model
  • adding layers to the model
  • compiling the model

Part 4: Training the Model

  • fetching batches of data
  • training, testing and evaluating the model
  • plotting graphs of the model loss and accuracy during training

Part 5: Predicting Samples

  • fetching batches of samples
  • predicting fetched samples

Part 6: Drawing Doodles

  • creating a new Painter class to allow users to draw their own doodles with the mouse
  • defining painting objects: drawing area, bitmaps, pencil
  • adding a function for drawing a smooth line between two points using quadratic curves

Part 7: Recognizing Doodles

  • resizing doodle drawing to the required size of 28x28
  • normalizing array of pixels before passing it as the input of the CNN model
  • predicting doodle

Part 8: Adding More Doodle Categories

  • adding 10 doodle categories