🤖 Gesture Recognition System

An advanced computer vision system for comparing and analyzing hand gestures using MediaPipe and OpenCV.

🎯 Overview

This project provides a sophisticated gesture recognition system that can:

• Record reference gestures from video files
• Capture live gestures using your webcam
• Compare gestures using advanced similarity algorithms
• Provide detailed scoring and quality assessment

✨ Features

• 🎥 Real-time Gesture Capture: Live webcam gesture recording
• 📹 Video Processing: Extract hand keypoints from reference videos
• 🔍 Advanced Comparison: Cosine similarity-based gesture matching
• 📊 Intelligent Scoring: Aggressive scoring algorithm for precise matching
• 🎯 Quality Assessment: Qualitative feedback on gesture similarity
• 🤖 MediaPipe Integration: Leverages Google's MediaPipe for accurate hand tracking
• 📈 Frame-by-Frame Analysis: Detailed comparison across multiple video frames

🧠 How It Works

1. Keypoint Extraction: Uses MediaPipe to extract 21 hand landmarks per hand
2. Feature Vector: Creates 126-dimensional feature vectors (21 points × 3 coordinates × 2 hands)
3. Similarity Calculation: Computes cosine similarity between reference and live gestures
4. Scoring Algorithm: Applies aggressive penalty for medium-quality matches to ensure precision

📊 Scoring System

Score Range	Assessment	Description
90-100%	Excellent	Near-perfect gesture match
70-89%	Good	Strong similarity with minor differences
50-69%	Fair	Moderate similarity
30-49%	Poor	Significant differences
0-29%	Failed	Gestures don't match

🛠️ Installation

Prerequisites

• Python 3.8+
• PyTorch
• OpenCV
• MediaPipe
• NumPy

Setup

1. Clone the repository

   git clone https://github.com/yourusername/gesture-recognition.git
   cd gesture-recognition

2. Install dependencies

   pip install opencv-python mediapipe numpy

3. Prepare your reference video

   • Place your reference gesture video as reference_video.mp4 in the project directory
   • Ensure clear hand visibility and good lighting

🚀 Quick Start

1. Prepare reference video

   • Place your reference gesture video as reference_video.mp4
   • Ensure clear hand visibility

2. Run the gesture comparison

   python gesture_recognition.py

3. Follow the interactive process

   • The system will automatically process your reference video
   • Your webcam will activate for live gesture recording
   • Perform your gesture in front of the camera
   • Press 'q' to stop recording and see the similarity results

� Example Output

Frames collected: 45 (reference), 38 (live)
Raw Similarity Score: 87.34%
Final Score: 42.15%
Assessment: Fair match

🔧 Configuration

Edit gesture_recognition.py to adjust:

• MIN_FRAMES_REQUIRED: Minimum frames for valid gesture (default: 10)
• FPS_REDUCTION_FACTOR: Frame processing rate (default: 2)
• FIXED_KEYPOINTS: Number of keypoints per gesture (default: 126)

Scoring Algorithm Customization

The apply_aggressive_scoring() function can be modified to change how similarity scores are transformed:

def apply_aggressive_scoring(normalized_similarity):
    # Modify these ranges to adjust scoring sensitivity
    if normalized_similarity >= 0.90:
        return normalized_similarity * 100  # Keep high scores
    elif 0.80 <= normalized_similarity < 0.90:
        # Aggressive penalty for medium scores
        return (0.20 + (normalized_similarity - 0.80) * 3) * 100
    # ... continue customization

🤝 Contributing

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

🤝 Contributing

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

🙏 Acknowledgments

• MediaPipe: For excellent hand tracking capabilities
• OpenCV: For computer vision processing
• NumPy: For numerical computations

---

Built with ❤️ by Rithwik