Advancing Workplace Safety: A Deep Learning Approach for Personal Protective Equipment Detection using Single Shot Detector

Personal Protective Equipment (PPE) serve as crucial components for engineers in the construction industry and other sectors, ensuring the safety and well-being of workers. An effective object detection system plays a pivotal role in identifying and monitoring the usage of PPE. This research focuses on testing a detection system utilizing a dataset comprising 132 scenarios from construction sites. The results indicate the system’s successful detection of PPE with a high level of confidence. Evaluation metrics such as mAP50(B), mAP50-95(B), precision(B), and recall(B) were employed to assess system performance. The testing yielded mAP50(B) of 0.768, mAP50-95(B) of 0.516, precision(B) of 0.831, and recall(B) of 0.693. These outcomes demonstrate that the PPE detection system, utilizing the SSD algorithm, exhibits satisfactory performance in recognizing and predicting PPE objects. However, there is room for improvement and further development, especially in addressing challenges such as variations in image capture, scale, and more complex environmental conditions. This study makes a significant contribution to the field of PPE detection and can serve as a foundation for the advancement of object detection systems. A reliable detection system holds immense potential in enhancing safety and occupational health across diverse industrial sectors.