hn8888/yolo-light repository overview

⁠🚀 Lightweight YOLO Object Detection API for Raspberry Pi 4

Real-time object detection API built with FastAPI and YOLOv5n, optimized for Raspberry Pi 4 and low-resource environments.

⁠✨ Key Features

• ⚡ Ultralight: YOLOv5n model (7.5MB) with 80 COCO classes detection
• 🏗️ Multi-Architecture: Supports amd64 and arm64 (Raspberry Pi 4+)
• 🔧 Flexible Models: Easily switch between YOLOv5/YOLOv11 variants via MODEL_NAME environment variable
• 🌐 REST API: 4 endpoints with JSON and visual output modes
• 💾 Efficient: 800MB-1.2GB runtime memory on RPi4
• 🐳 Docker Native: Automated GitHub Actions multi-architecture builds

⁠🎯 Quick Start

docker run -d -p 8000:8000 --memory=1.5G hn8888/yolo-light:arm64 curl http://localhost:8000/health⁠

⁠📚 Endpoints

• ✅ GET /health: API and model status
• ✅ POST /detect: JSON response with detections
• ✅ POST /detect-visual: PNG image with bounding boxes drawn
• ✅ GET /: API information

⁠🔨 Environment Variables

• MODEL_NAME (default: yolov5n.pt): Choose any YOLO model
• PORT (default: 8000): API port

⁠🎨 Available Models

yolov5n, yolov5s, yolov5m, yolov5l, yolov5x, yolov11n, yolov11s, yolov11m, yolov11l, yolov11x

⁠📋 Requirements

• 🖥️ Raspberry Pi 4 (2GB+ RAM recommended)
• 🐳 Docker 20.10+
• 💿 1.5-2GB disk space for image

⁠📜 License

CC BY-NC 4.0 (Non-Commercial Use Only)

⁠📖 Documentation

For complete guides, deployment instructions, and examples, visit: https://github.com/hlavrencic/yolo-light⁠

---

🎉 Optimized for RPi4 and low-power environments!