Research on Safety Helmet Wearing Detection Algorithm Based on Lightweight Convolution and Spatial Attention

Aiming at the problems of low detection accuracy and missed detection caused by low resolution, small target and partial occlusion in the safety helmet wearing detection task, in order to reduce the occurrence of violations and accidents in industrial production, an object detection algorithm integrating lightweight convolution and spatial attention module combining coordinates is studied, which is embeds the Space-to-depth-convolution (SPD-conv) and coordinate attention module (CA) on the basis of yolov5 (6.0), and replaces the loss function with a SIoU function that converges faster. The experimental results show that compared with the original network algorithm, the article algorithm has a little improvement in detection accuracy, and has a better detection effect on small and low resolution targets.