AirCanvas

AirCanvas is a computer vision project that lets you draw using hand gestures. Built with OpenCV and MediaPipe, it tracks your hand movements to simulate basic painting—like MS Paint, but without a mouse.

About

AirCanvas transforms your webcam into an interactive digital canvas, allowing you to create art using just your hand gestures in the air. The application uses:

• MediaPipe for real-time hand tracking and gesture recognition
• OpenCV for image processing and drawing operations
• Computer Vision techniques to interpret hand movements as drawing commands

The project demonstrates how computer vision can be used to create natural human-computer interfaces, enabling creative expression without physical input devices.

Installation

Prerequisites

• Python 3.8 or higher
• Webcam or camera

Setup

1. Clone the repository:

   git clone https://github.com/d4min/aircanvas.git
   cd aircanvas
   

2. Create a virtual environment (recommended):

   # Using conda
   conda create -n aircanvas python=3.8
   conda activate aircanvas
   
   # Or using venv
   python -m venv aircanvas
   source aircanvas/bin/activate  # On Windows: aircanvas\Scripts\activate
   

3. Install the required packages:

   pip install -r requirements.txt
   

4. Run the application:

   python src/main.py
   

Usage

AirCanvas recognises different hand gestures to control drawing functions:

Drawing Gesture

Make a pinch gesture with your index finger and thumb to draw. Move your hand while maintaining the pinch to create lines and shapes.

Select Gesture

Point with your index finger to select colors from the palette on the right side of the screen.

Erase Gesture

Show an open palm to erase. A circle will appear around your index finger showing the eraser size.

Controls and Features

• Colour Palette: Located on the right side of the screen
• Current Colour: Displayed in the top-left corner
• Exit: Press 'q' to quit the application

Audio Recognition

AirCanvas includes an audio recognition feature that allows you to control the application using voice commands. This feature leverages the speech_recognition library to interpret spoken words and execute corresponding actions within the application.

Supported Commands

• "Clear": Clears the canvas.
• "Exit": Closes the application.
• "Blue", "Red", "Green", "Yellow", "White": Changes the drawing color to the specified color.

The audio recognition runs in a separate thread, allowing it to listen for commands continuously while you draw.

Contributor Acknowledgment

Special thanks to @AlexCaranha for implementing the speech recognition functionality.

Implementation Progress

Phase 1: Setup ✅

• [x] Project structure setup
• [x] Dependencies installation
• [x] Camera integration
• [x] Basic configuration

Phase 2: Hand Tracking ✅

• [x] MediaPipe integration
• [x] Landmark detection
• [x] Coordinate system
• [x] Real-time tracking

Phase 3: Gesture System ✅

• [x] Basic gesture detection
• [x] Gesture states
• [x] Action mapping
• [x] Gesture refinement

Phase 4: Canvas System ✅

• [x] Drawing implementation
• [x] Color system

Phase 5: Features 🔄

• [x] Audio commands
• [ ] Tool panel
• [ ] Save/Load system
• [ ] Shapes e.g. Rectangle, Circle
• [ ] Pen thickness

Legend

✅ Complete 🔄 In Progress ⏳ Planned