Key points are not available for this paper at this time.
The detection of personal protective equipment (PPE) has grown to be a significant concern in a variety of industries. Ensuring that workers have the proper safety gear, such as hard helmets, gloves, eye protection, and so forth, it aims to reduce accident rates and fatalities. Through the years, the majority of PPE identification has been accomplished through human inspection and vision-based procedures. On the other hand, human capabilities limit manual examination, while vision-based systems frequently struggle to recognize items effectively in low light or at a distance. On the other hand, the utilization of the YOLO (You Only Look Once) algorithm has demonstrated superiority concerning speed, effectiveness, and accuracy. As a result, we use the YOLO algorithm in this study to present an efficient vision-based PPE detection system, where the YOLOv8 large gave the highest average precision of 91.7%. The training procedure involved utilizing the widely recognized and accessible CHV datasets. The datasets were trained using YOLOv8 models, and then investigated and compared from five different perspectives: PPE identification average precision, recall, F1 score and Training speed.
Mohona et al. (Thu,) studied this question.