Extract image from streaming and detect specific patterns in the image.
Description:
- Extract image from streaming, apply post-processing to the image then send to the recognition service.
- Generate dataset and train YOLOv4 model, optimize detection confidence up to 90%.
- Build a WebSocket service for keeping push results to the frontend.
- Setup Nginx as a reverse proxy to handle WebSocket routing and static files.
- Containerize the above services for deployment on GPU machines quickly.
Tech Stack:
- WebSocket server: gorilla/webscoket
- Reverse proxy: Nginx
- Machine learning model: YOLOv4
- Streaming to Image: FFmpeg
- Service communication: ZeroMQ