Multi-Camera Real-Time Object Detection & ROI Event Tracking System
Built a real-time multi-camera object detection system using YOLOv7, processing four simultaneous video streams with custom-trained models across seven object classes: Car, Bike, Helmet, Jacket, Fire, Smoke, and Person. Architected a scalable message-driven pipeline using RabbitMQ and Redis for high-volume frame processing. Designed ROI-based alerting to detect and save frames when objects crossed predefined zones.