Headgear Compliance in Development Sectors Through Deep Learning Algorithms

The paper introduces a comprehensive approach to enhance safety practices in high-risk professions, specifically focusing on headgear compliance in the transportation and construction sectors. Leveraging the advancements of the You Only Look Once (YOLO) algorithm, particularly the YOLOv8-L model, the study presents a novel helmet and vest detection system for real-time monitoring of personal protective equipment (PPE). The YOLOv8-L model, with its compact size of 21 MB and rapid single epoch run-time of 11 seconds, achieved an impressive mean average precision (mAP) accuracy rate of 96.8%. The system, trained on a meticulously curated dataset encompassing various lighting conditions, camera angles, and helmet types, highlights robust performance, attaining a recall of 0.99 and an average precision of 0.99. The implementation of this deep learning algorithm not only ensures efficient and reliable detection of safety helmets but also significantly contributes to improving safety measures in hazardous work environments, thereby promoting headgear compliance, and reducing accidents in the transportation and construction sectors.