Answers are grounded in this paper and cite their sources.
Construction sites are highly dynamic and inherently hazardous environments where safety compliance is crucial yet challenging to enforce manually. Failing to wear essential personal protective equipment, such as helmets and safety vests, and unauthorized entry into restricted risk zones are major contributors to preventable injuries. This study introduces an intelligent automated safety monitoring system based on computer vision and deep learning. The proposed framework integrates the YOLOv8 object detection model with facial recognition to perform automated worker identification, real-time PPE compliance verification, and restricted zone monitoring. When a safety violation is detected, the system captures visual evidence, triggers localized alerts, and records metadata including worker identity, timestamp, location, and violation type. To support worker identification under varying environmental conditions, the framework employs a video-based enrollment process combined with attendance-based database filtering. The system was evaluated through a three-day field deployment at a real-world construction site in Egypt, demonstrating reliable performance under changing ambient conditions. The results demonstrate the feasibility of continuous automated safety monitoring under real-world construction-site conditions and show the framework’s ability to detect, document, and associate safety violations with structured incident records. The proposed framework offers an AI-driven approach to support workplace safety monitoring and contractual compliance review.
No takes yet. Share an insight, caveat, or question.
Shawky et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: