Safety Helmet Detection Algorithm Based on Improved YOLOv7

The wearing of helmets is crucial for construction sites, which can protect workers from injuries caused by head impacts and reduce the occurrence of safety accidents. The image size captured by the construction site camera is small, and it is inevitable that there are omissions and wrong detections; considering the complexity of the construction site, there are also the problems of slow detection speed and insufficient detection accuracy, so this paper proposes a helmet detection algorithm based on the improved YOLOv7. The simam module is added to the backbone network of YOLOv7, which makes the model pay more attention to the extraction of main features and reduces the problems of omission and misdetection; based on the NWD method, the loss function is improved to improve the regression accuracy and accelerate the convergence. The experimental results show that the improved algorithm has better results than the original YOLOv7 model by improving the accuracy by 2.1% and mAP by 1.4%.