Doodle Classifier

Interactive web application to classify user-drawn doodles into one of ten categories using a neural network.

Table of Contents

• Overview
• Features
• Setup Instructions
  • Prerequisites
  • Running the Application

Overview

Figure 1: Demonstration of the Doodle Classifier in action, showing how users can draw, classify, and clear their doodles.

The Doodle Classifier is a Flask-based web application that enables users to draw doodles on a canvas and classify them into one of the following ten categories:

• butterfly, coffee cup, drums, frog, hamburger, leaf, microphone, onion, pizza, wristwatch

Using a pre-trained neural network, the application processes user-drawn images and identifies the most likely category. The user can clear the canvas or redraw for further experimentation.

Features

• Drawing Interface: Users can freely draw in the browser-based square canvas.
• Doodle Classification: Classifies drawn images into one of the ten categories with a trained neural network.
• Real-time Interactivity: Quick feedback with the ability to clear the canvas and try again.

Setup Instructions

Prerequisites

To run this project, you need:

• Python 3.8+
• Recommended to use a virtual environment (venv) to manage dependencies

Running the Application

1. Clone the repository:

git clone https://github.com/vulong2505/Doodle-Classifier.git
cd doodle-classifier

2. Install dependencies:

pip install -r requirements.txt

3. Start the application:

py app/server.py

4. In the terminal, open the ULR shown in the terminal (it might be http://127.0.0.1:5000/). Refer to the image below to find the link after starting the server.