Drone Testing in Action
Shown here is a real-time test of the CAPS system using a DJI Mini 4K drone. The environment simulates construction PPE compliance monitoring, where violations (like missing hardhats or vests) trigger immediate alerts.
The Construction Accident Prediction System (CAPS) is an innovative AI-powered safety solution designed to monitor and detect real-time safety violations on construction sites using computer vision and drone technology. CAPS combines cutting-edge deep learning algorithms, object detection models, and streaming integration to help construction managers and safety officers prevent accidents before they occur.
Shown here is a real-time test of the CAPS system using a DJI Mini 4K drone. The environment simulates construction PPE compliance monitoring, where violations (like missing hardhats or vests) trigger immediate alerts.
CAPS identifies personal protective equipment (PPE) compliance, such as the presence of hardhats, safety vests, and face masks, and tracks objects like persons, machinery, and vehicles. When a safety violation is detected, the system triggers a visual and audible alert.
Construction sites are among the most hazardous work environments globally. This project was inspired by the need for a proactive safety system that could scale across large sites, operate continuously, and instantly detect and report breaches.
RTMP Latency: Solved via FFmpeg buffer tuning and throttling frames to improve performance.
FFmpeg Integration: Used image2pipe with NumPy + OpenCV and frame.copy() to resolve memory issues.
Model Optimization: Used yolov8n, resized frames, and async alerts to ensure smooth operation.
🏆 First Place & Gold Award for Innovation and Impact (University of Bedfordshire Project Exhibition 2025)
CAPS is more than a project – it's a step toward proactive, data-driven safety in construction.
Drone testing for PPE violation detection
CAPS dashboard live detection display
Integrated weather alert with detection overlay