Safety Helmet Wearing Detection Algorithm Based on Improved Yolov5 s

In view of the current detection requirements for personal protective equipment safety helmets on the construction site, the existing algorithms have problems such as low detection accuracy, slow speed, missed detection and false detection in complex scenes. This paper proposes a safety helmet wearing detection algorithm based on improved Yolov5 s, which effectively improves the detection accuracy under the premise of ensuring the detection speed. Firstly, CBAM attention module is added to the backbone network to make the model pay more attention to the defect area of gear parts and improve the recognition ability of the model. Secondly, the BiFPN (Bidirectional Feature Pyramid) is combined in the neck network, and the PANet in the Yolov5 benchmark structure is replaced by the BIFPN structure, and the bidirectional cross-scale connection is used to enhance the feature representation ability. Finally, the CIoU loss function is replaced by the SIoU loss function to improve the accuracy of detection. The experimental results show that the average accuracy value (mAP @ 0.5) of the improved Yolov5 s model for helmet wearing detection reaches 98.8 %, which is 2.3 % higher than that before the model improvement, and can meet the requirements of real environment use.