A Personal Protective Equipment Safety Monitoring System in Industrial Scenes

Personal protective equipment (PPE) is required in many industrial workplaces to ensure worker safety. This paper proposes a design of a PPE safety monitoring system (PPE-SMS) that analyzes live video streams from IP cameras to determine whether workers are wearing PPE. Firstly, a method of PPE wearing judgment based on deep learning object detection and geometric correlation analysis is presented. Secondly, a design for an efficient video analysis pipeline that includes the PPE wearing judgment method is proposed. Thirdly, a PPE-SMS is developed using the designed pipeline. Following evaluation, the PPE wearing judgment method achieves 0.93 precision and 0.96 recall; the video analysis pipeline achieves an average total latency of 241.57ms, and the PPE-SMS functions as expected. The results showed that the system can meet the monitoring requirements of real-world industrial scenarios.