The retrieval system of wearing safety helmet based on deep learning

摘要

#$%^&*AU2020100711A420200611.pdf#####ABSTRACT Head injury of construction workers is an important cause of building casualties. Wearing safety helmet is an effective measure to prevent brain injury accidents of construction workers, while unsafe behaviors of workers who do not wear safety helmet often occur in actual work. Therefore, the target detection of construction workers wearing helmets will provide a new perspective for in-depth recognition and active prevention of safety accidents. The traditional construction site has a series of problems, such as low level of safety management, small scope of management, mainly relying on the subjective monitoring of safety management personnel and poor timeliness, unable to monitor the whole process and so on. In view of the above situation, a Yolo method based on pytorch framework, which has good generalization performance and fast performance, is proposed to detect the wearing status of workers' safety helmets. The specific implementation steps are as follows: firstly, a large number of images of safety helmets are obtained through the web crawler as the initial recognition library, and then image processing is carried out. 1Figure 4 Figure 5 *~ 26x~26-255 416-416-3 DBL*5 S5252255 DBL*5 52-52x~255 Darknetconv2DBNLeaky Res unit Resblackbody Res-unit*n Figure 6 2