Abstract: As a protective equipment to prevent head injuries,safety helmets are required to be worn when entering high-risk places such as power plants. In actual work,it happens from time to time to enter the work site without wearing a helmet. To solve this problem,it proposes an intelligent detection technology based on safety helmets. This technology uses the YOLOv5 algorithm to train the data,and uses the YOLOv5s model with the smallest network depth and width in the YOLOv5 series. The test results show that the average precision of training and testing in self-collected data is 95. 4%,which can meet the requirementof real-time monitoring of people wearing safety helmets in high-risk places such as power plants.