Full-Stack Laboratory PPE Detection Using Deep Learning
Real-time video processing system with sub-second inference latency using cascading YOLO models and distributed microservices architecture.
Project Details
Developed real-time video processing system using FastAPI, async/await patterns, and cascading YOLO models with multi-stage detection pipelines, achieving sub-second inference latency through optimized OpenCV stream processing.
Architected scalable edge computing backend with RBAC, JWT authentication via Supabase, RESTful APIs, and microservices architecture including violation detection engine, cloud sync service, and automated email reporting.
Implemented distributed system components with intelligent stream reconnection, IoU-based person tracking algorithms, violation debouncing mechanisms, and real-time database synchronization.
Key Features
- Developed real-time video processing with async/await patterns and cascading YOLO models
- Architected scalable edge computing backend with RBAC, JWT authentication, and microservices
- Implemented distributed system components with intelligent stream reconnection and violation detection