Holly Stream

A real-time video streaming application with integrated AI-powered object detection. Stream your webcam feed with live bounding box detection via RTMP/HLS protocols to media players, web browsers, or remote servers.

---

Overview

Holly Stream captures video from your camera, applies YOLO-based object detection with visual bounding boxes, and streams the processed feed in real-time. The application can also function as a simple live stream camera when object detection is disabled, making it ideal for security systems, monitoring applications, and AI-powered surveillance.

Key Features

• Real-time Object Detection: YOLOv5/v11 models with 80 COCO dataset classes
• GPU Acceleration: Optimized for NVIDIA Jetson and CUDA-enabled devices using TensorRT
• Flexible Streaming: RTMP protocol for media players and HLS for web browsers
• Customizable Detection: Configurable classes, confidence thresholds, and detection parameters
• Multiple Deployment Options: Local, LAN, or remote web server streaming
• Docker Support: Fully containerized with Docker Compose orchestration
• Multi-Platform: Dedicated branches for Jetson, Linux, and Raspberry Pi architectures
• Custom Model Support: Train and deploy your own YOLOv5 models

Use Cases

• Home security and surveillance systems
• Wildlife monitoring and detection
• Smart doorbell and entry monitoring
• Pet detection and tracking
• Traffic and parking monitoring
• Custom object detection applications

Requirements

• Docker and Docker Compose
• NVIDIA GPU (for GPU-accelerated inference)
• Webcam or CSI camera
• 4GB+ RAM (8GB recommended for Jetson devices)

Platform-Specific Branches

This project maintains three platform-optimized branches, each tailored for specific hardware architectures:

Branch	Platform	Tested Hardware	Camera Support
jetson	NVIDIA Jetson	Jetson Nano (JetPack 4.6.4)	CSI (IMX219-160 8MP)
linux	Linux x86_64	Ubuntu 22.04 + GTX 1060 6GB	USB (Logitech 1080p)
raspbian	Raspberry Pi	Raspberry Pi 4 (Debian Bookworm)	CSI (IMX519 16MP)

Getting Started

Clone the branch corresponding to your target platform:

For NVIDIA Jetson Devices:

# HTTPS
git clone --branch jetson --depth 1 https://github.com/rcland12/holly-stream.git

# SSH
git clone --branch jetson --depth 1 [email protected]:rcland12/holly-stream.git

For Linux Systems with NVIDIA GPU:

# HTTPS
git clone --branch linux --depth 1 https://github.com/rcland12/holly-stream.git

# SSH
git clone --branch linux --depth 1 [email protected]:rcland12/holly-stream.git

For Raspberry Pi:

# HTTPS
git clone --branch raspbian --depth 1 https://github.com/rcland12/holly-stream.git

# SSH
git clone --branch raspbian --depth 1 [email protected]:rcland12/holly-stream.git

After cloning, navigate to the branch README for detailed installation and deployment instructions specific to your platform.

Quick Start

1. Clone the appropriate branch for your platform
2. Create an .env file with your configuration
3. Run ./run.sh to start streaming
4. View the stream via media player (VLC, OBS) or web browser

For comprehensive setup instructions, prerequisites, and configuration options, refer to the README in your selected branch.

Technologies

• Computer Vision: OpenCV, YOLOv5/v11, TensorRT
• Deep Learning: PyTorch, TorchVision, ONNX
• Inference: NVIDIA Triton Inference Server
• Streaming: FFmpeg, Nginx with RTMP module, HLS
• Containerization: Docker, Docker Compose

Contributing

Contributions are welcome! Please see CONTRIBUTING.md for detailed guidelines on how to contribute to this project.

Areas for Improvement

• Additional platform and architecture support
• Streaming latency optimization (WebRTC, DASH, UDP protocols)
• Enhanced deployment methods for non-Docker environments
• Performance optimizations and code refactoring
• Documentation improvements and tutorials

License

This project is licensed under the terms specified in the LICENSE file.

Support

For issues, questions, or feature requests, please open an issue on the GitHub Issues page.

---

Note: Each platform branch contains detailed documentation specific to that architecture. Always refer to the branch-specific README for installation steps, prerequisites, and deployment instructions.