Fusion Nvr

A custom NVR (Network Video Recorder) solution based on Frigate, featuring object detection, tracking, and recording capabilities. This project includes custom branding, Docker deployment, and optimized configuration for production use.

🌟 Features

• Object Detection: YOLOv8-based detection with ONNX runtime
• Object Tracking: Multi-object tracking with configurable methods
• Video Recording: Automatic event-based recording with retention policies
• Web Interface: Modern React-based web UI with custom branding
• Docker Deployment: Easy deployment with Docker Compose
• Cross-Platform: Works on Windows, Linux, and macOS
• Real-time Processing: Live camera feeds and real-time detection

🚀 Quick Start

Docker Deployment (Recommended)

Prerequisites:

• Docker Desktop installed and running
• At least 4GB RAM available for Docker

1. Start Docker Desktop (if not already running)

2. Build the frontend:

cd web
npm install
npm run build:local
cd ..

3. Start Fusion Nvr:

docker compose up -d

4. Access the web interface:

• Open your browser and navigate to: http://localhost:5001
• The web interface will show "Fusion Nvr" with custom branding

5. Stop Fusion Nvr:

docker compose down

Python Environment Setup (Alternative)

For Python 3.11+ (Latest):

setup-python-env.bat

For Python 3.10 (Compatible):

setup-python310-env.bat

Run Fusion Nvr:

# For Python 3.11+
.\run-frigate-python.bat

# For Python 3.10
run-frigate-python310.bat

Manual Setup

1. Create Virtual Environment:

python -m venv venv
venv\Scripts\activate.bat

2. Install Dependencies:

# Install PyTorch (CPU version)
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cpu

# Install DeepOCSORT and tracking packages
pip install git+https://github.com/GerardMaggiolino/Deep-OC-SORT.git
pip install torchreid lap cython-bbox

# Install Frigate requirements
pip install -r requirements-python.txt

3. Run Frigate:

python -m frigate

📁 Project Structure

frigate/
├── frigate/                 # Main application code
├── web/                     # React web interface (frontend)
│   ├── src/                # React source code
│   ├── dist/               # Built frontend (generated)
│   └── images/             # Custom logo and images
├── config/                 # Configuration files
│   └── config.yml          # Main configuration file
├── input/                  # Video input files
├── docker-compose.yml       # Docker Compose configuration
├── requirements-python.txt # Python dependencies
├── setup-python-env.bat    # Python 3.11+ setup script
├── setup-python310-env.bat # Python 3.10 setup script
├── run-frigate-python.bat  # Run Fusion Nvr (Python 3.11+)
└── run-frigate-python310.bat # Run Fusion Nvr (Python 3.10)

⚙️ Configuration

Basic Configuration

Edit config/config.yml to configure cameras, detection, and recording:

# MQTT (optional, disabled by default)
mqtt:
  enabled: false

# Global model configuration
model:
  model_type: yolo-generic
  width: 640
  height: 640
  input_tensor: nchw
  input_dtype: float
  input_pixel_format: bgr
  path: "/config/yolo8n.onnx"  # Path to YOLOv8 ONNX model

# Detectors
detectors:
  onnx:
    type: onnx
    device: AUTO  # AUTO, CPU, or GPU (if available)

# Cameras
cameras:
  test_camera:
    ffmpeg:
      inputs:
        - path: input/test_video.mp4
          roles:
            - detect
            - record
    detect:
      width: 1280
      height: 720
      fps: 5
    record:
      enabled: true
      events:
        retain:
          default: 30
          mode: motion
          objects:
            person: 30
    live:
      height: 720
      quality: 8

# Object tracking
track:
  enabled: true
  max_disappeared: 30
  max_distance: 50

Model Setup

YOLOv8 Model:

1. Export YOLOv8 to ONNX format:

   from ultralytics import YOLO
   model = YOLO("yolov8n.pt")
   model.export(format="onnx", imgsz=[640, 640], opset=20, dynamic=True, dynamo=True)

2. Place the exported yolov8n.onnx file in the config/ directory
3. Update the model.path in config.yml to point to the model file

Docker Configuration

The docker-compose.yml file includes:

• Volume mounts for config, recordings, and web interface
• Named volumes for database and model cache (avoids Windows filesystem issues)
• Port mappings for web UI (5001), API (5000), and RTSP (8554, 8555)
• Environment variables for timezone and RTSP password

🔧 Requirements

Docker Deployment

• Docker Desktop: Latest version
• RAM: 4GB minimum, 8GB recommended for Docker
• Storage: 10GB free space
• OS: Windows 10/11, Linux, macOS

Python Deployment (Alternative)

• Python: 3.10 or 3.11+ (3.11+ recommended)
• RAM: 4GB minimum, 8GB recommended
• Storage: 10GB free space
• OS: Windows 10/11, Linux, macOS

Python Packages (for Python deployment)

• PyTorch: 2.1.0+ (CPU version)
• OpenCV: 4.8.0+
• NumPy: 1.24.0+
• ONNX Runtime: 1.15.0+
• Frigate: Latest version

🐛 Troubleshooting

Docker Issues

1. Docker Desktop not running:

• Ensure Docker Desktop is started before running docker compose up -d
• Check Docker Desktop status in system tray

2. Port already in use:

• Change ports in docker-compose.yml if 5000, 5001, or 8554 are in use
• Check for other services using these ports

3. Web interface shows old images:

• Rebuild the frontend: cd web && npm run build:local && cd ..
• Restart Docker: docker compose down && docker compose up -d
• Clear browser cache (Ctrl+Shift+R)

4. Database I/O errors:

• The Docker setup uses named volumes to avoid Windows filesystem issues
• If issues persist, check Docker volume permissions

5. Model not found:

• Ensure yolo8n.onnx is in the config/ directory
• Check the model.path in config.yml matches the file location

Python Deployment Issues

1. Missing packages:

pip install -r requirements-python.txt

2. NumPy compatibility issues:

pip install "numpy<2.0"

3. ONNX Runtime installation:

pip install onnxruntime>=1.15.0

4. PyTorch installation fails:

pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cpu

Python Version Issues

For Python 3.10:

• Use setup-python310-env.bat
• May need older package versions

For Python 3.11+:

• Use setup-python-env.bat
• Full feature support

📚 Documentation

• Frigate: Official Documentation
• YOLOv8: Ultralytics Documentation
• Docker: Docker Documentation
• ONNX Runtime: ONNX Runtime Documentation

🤝 Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Test thoroughly
5. Submit a pull request

📄 License

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

🙏 Acknowledgments

• Frigate Team: For the excellent NVR software foundation
• Ultralytics: For YOLOv8 object detection models
• ONNX Runtime Team: For the efficient inference engine
• React Community: For the web framework
• Docker Team: For containerization platform

📞 Support

If you encounter issues:

1. Check the Troubleshooting section
2. Search existing Issues
3. Create a new issue with detailed information
4. Join the community discussions

---

🎨 Customization

Changing the Logo

1. Replace web/images/logo.png with your custom logo (recommended: 512x512px PNG)
2. Rebuild the frontend: cd web && npm run build:local && cd ..
3. Restart Docker: docker compose down && docker compose up -d

Changing the Application Name

1. Update translation files in web/public/locales/en/ (replace "Fusion Nvr" with your name)
2. Update web/site.webmanifest with your application name
3. Rebuild the frontend and restart Docker

---

Happy Monitoring! 🎯