ObjectDetection

A real-time object detection system that combines YOLO (You Only Look Once) and Faster R-CNN for enhanced detection accuracy. This project demonstrates the power of ensemble methods in computer vision by merging predictions from two state-of-the-art object detection models.

🚀 Features

• Hybrid Detection: Combines YOLO and Faster R-CNN predictions for improved accuracy
• Real-time Processing: Live camera feed with real-time object detection
• Multiple Views: Three display windows showing different detection methods
• Smart Merging: Uses IoU (Intersection over Union) to merge overlapping detections
• Non-Maximum Suppression: Eliminates duplicate detections for cleaner results
• Interactive Interface: Press 'q' to quit the application

🎯 Detection Methods

1. YOLO (You Only Look Once)

• Color: Green bounding boxes
• Speed: Fast inference for real-time applications
• Accuracy: Good for general object detection

2. Faster R-CNN

• Color: Blue bounding boxes
• Speed: Slower but more accurate
• Accuracy: Excellent for precise object localization

3. Hybrid Approach

• Color: Red bounding boxes
• Method: Combines both models using IoU thresholding
• Result: Best of both worlds - speed and accuracy

🛠️ Requirements

• Python 3.8+
• OpenCV
• PyTorch
• Ultralytics
• Torchvision

📦 Installation

1. Clone the repository

   git clone https://github.com/divyanshsingh07/ObjectDetection.git
   cd ObjectDetection

2. Create virtual environment

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies

   pip install torch torchvision ultralytics opencv-python

🚀 Usage

Real-time Camera Detection

cd Code/yolo_project
source venv_new/bin/activate  # Use existing virtual environment
python main.py

Alternative Version

cd Code/Custom
python main.py

🎮 Controls

• Press 'q': Quit the application
• Camera: Automatically uses default camera (index 0)

📁 Project Structure

Code/
├── yolo_project/          # Main project directory
│   ├── main.py           # Main application with camera access
│   ├── initial.py        # Initial implementation
│   ├── venv_new/         # Virtual environment
│   └── yolov5su.pt       # YOLO model weights
├── Custom/                # Alternative implementation
│   ├── main.py           # Custom version with camera access
│   └── initial.py        # Initial custom implementation
├── data/                  # Training data (ignored by git)
├── vehicle dataset/       # Vehicle detection dataset (ignored by git)
└── .gitignore            # Git ignore rules

🔧 Technical Details

Model Architecture

• YOLO: Single-stage detector for real-time performance
• Faster R-CNN: Two-stage detector for high accuracy
• Ensemble: Combines predictions using IoU-based merging

Performance Optimization

• GPU Acceleration: Supports CUDA for faster inference
• Batch Processing: Efficient frame processing pipeline
• Memory Management: Proper cleanup of OpenCV windows

📊 Supported Objects

The system can detect various objects including:

• People
• Vehicles (cars, trucks, buses)
• Animals
• Common objects (chairs, tables, etc.)
• And many more COCO dataset classes

🤝 Contributing

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

📝 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• Ultralytics for YOLO implementation
• PyTorch for deep learning framework
• OpenCV for computer vision operations

📞 Contact

• Author: Divyansh Singh
• GitHub: @divyanshsingh07
• Project: ObjectDetection

---

⭐ Star this repository if you find it helpful!