AI System Targets Real-Time Safety Helmet Compliance Monitoring
Researchers developed a system that combines helmet detection with head-location technology to determine whether workers are actually wearing protective headgear.
- By Danielle Naidu
- Oct 05, 2026
Researchers have developed an artificial intelligence system designed to improve real-time monitoring of safety helmet use in construction and industrial workplaces.
The system combines helmet detection with technology that identifies the location of a worker's head, allowing it to determine whether a detected helmet is actually being worn. The approach is intended to address a limitation of some automated PPE monitoring systems, which may detect a helmet in an image without reliably determining whether it is on a worker's head.
The research was published Oct. 5 in Scientific Reports. Read the study here.
The AI framework uses a YOLO11-based architecture to detect helmets and identify key points on a worker's body. It then compares the worker's head location with the detected helmet to determine whether they align.
Researchers evaluated the system using a construction PPE dataset. According to the study, the system achieved an F1 score of 0.961 and processed 110.7 frames per second. The researchers also reported that the approach reduced false compliance alarms by up to 67% compared with standalone YOLO detection models.
The system was designed to run on edge computing devices, processing information close to where video is collected rather than relying entirely on remote computing resources. Researchers also demonstrated the technology through a web-based platform that can monitor multiple live video streams and automatically report detected violations.
Researchers said the combination of accuracy and lower computing demands could make the approach suitable for scalable, real-time PPE compliance monitoring.
The study, "A lightweight pose-guided fusion approach for accurate safety helmet compliance monitoring on edge devices," was conducted by researchers from Chungbuk National University, Sookmyung Women's University and Heriot-Watt University Dubai.
About the Author
Danielle Naidu is assistant editor for Security Today, Campus Security Today, Occupational Health & Safety and Environmental Protection Online.