yihong1120/Construction-Hazard-Detection
Enhances construction site safety using YOLO for object detection, identifying hazards like workers without helmets or safety vests, and proximity to machinery or vehicles. HDBSCAN clusters safety cone coordinates to create monitored zones. Post-processing algorithms improve detection accuracy.
What it solves
Construction sites require constant monitoring to ensure worker safety and compliance. This project automates the detection of safety violations and hazardous proximity situations using live camera feeds, reducing the need for manual supervision.
How it works
The system processes live RTSP streams using YOLO-based computer vision models. It utilizes a distributed architecture with local YOLO worker processes that share GPU inference via shared memory. The runtime coordinates tasks using Redis and stores site configurations, user data, and violation records in PostgreSQL. Processed video is published via MediaMTX for HLS/WebRTC playback, and notifications can be sent via an FCM server.
Who it’s for
- Construction site managers and safety officers
- Industrial facility operators
- Developers building automated safety monitoring infrastructure
Highlights
- Detects missing hard hats, masks, or safety vests
- Identifies workers in restricted zones or near heavy machinery/vehicles
- Supports multiple languages for notifications (English, Chinese, French, etc.)
- Includes full model development utilities for training, evaluation, and data augmentation
- Provides a comprehensive suite of microservices for streaming, database management, and notifications