Enhancing Occupational Safety through Deep Learning-Based Helmet Detection in Hazardous Workplaces

Construction sites are among the most hazardous workplaces globally, where non-compliance with protective equipment standards contributes significantly to workplace accidents, including severe head injuries. Helmets, a crucial component of personal protective equipment (PPE), play a vital role in mitigating risks and preventing fatalities. However, manual monitoring of helmet compliance is often inefficient and prone to errors. Recent advancements in technology, particularly in the field of deep learning-based object detection, present promising solutions. This study focuses on the application of the YOLOv7 algorithm for real-time helmet detection. YOLOv7, a version of the You Only Look Once (YOLO) family, offers an optimal balance between speed and accuracy, making it well-suited for detecting small, moving objects such as helmets in dynamic construction environments. The model was trained on a diverse dataset that included various helmet types, colors, and environmental conditions, and was exposed to a wide range of lighting and weather scenarios. Performance metrics, including precision, recall, mAP, and F1 score, were used to assess the model, demonstrating its superior detection capabilities compared to its predecessor, YOLOv8. YOLOv7 achieved an effective balance between computational efficiency and detection accuracy, underscoring its potential to enhance occupational safety. The results also emphasize the critical importance of integrating automated systems to enforce safety regulations and minimize workplace hazards.