Professional Hand Tracking System

A real-time, production-style Hand Tracking System built with Python, OpenCV, and MediaPipe. This project demonstrates advanced computer vision techniques for gesture recognition, landmark stabilization, and interactive control.

🚀 Features

• Real-time Tracking: Detects and tracks 21 hand landmarks for multiple hands.
• Gesture Recognition: Identifies Open Palm, Fist, Pointing, Peace, and Pinch gestures.
• Landmark Stabilization: Smooths landmark jitter using exponential moving average for a premium experience.
• Clean UI Overlay: Minimalist design showing FPS, active gesture, and handedness.
• Bonus Interaction:
  • Mouse Control: Control your cursor using your index finger (Pointing gesture).
  • Virtual Drawing: Draw in mid-air using the Peace gesture.
  • Volume Feedback: Visual cues for volume control during Pinch gestures.

🛠️ Tech Stack

• Python: Core logic.
• OpenCV: Video capture and image processing.
• MediaPipe: State-of-the-art hand tracking models.
• PyAutoGUI: System-level interaction for mouse and volume control.
• NumPy: Numerical transformations and coordinate mapping.

📂 Project Structure

hand_tracking_system/
├── src/
│   ├── main.py           # Entry point & interaction logic
│   ├── tracker.py        # MediaPipe Hands wrapper
│   ├── recognizer.py     # Gesture analysis logic
│   ├── performance.py    # Stabilization and smoothing
│   └── utils.py          # Drawing and calculation helpers
├── requirements.txt      # Project dependencies
└── README.md             # Documentation

⚙️ Installation

1. Clone the repository:

   git clone https://github.com/krishanth7/ai-hand-tracking-system.git
   cd hand_tracking_system

2. Install dependencies:

   pip install -r requirements.txt

3. Run the application:

   python src/main.py

🎮 How to Use

• Show Palm: Displays "Open Palm".
• Make a Fist: Displays "Fist".
• Point Index Finger: Move the mouse cursor.
• Peace Sign (Index + Middle): Start drawing on the screen.
• Pinch (Thumb + Index): Triggers "Pinch" gesture (can be mapped to volume/zoom).
• Press 'C': Clear the virtual drawing.
• Press 'Q': Exit the application.

📝 Code Quality

This project follows Object-Oriented Programming (OOP) principles, with a focus on modularity, readability, and performance. Each component is isolated into its own module for easy maintenance and scalability.