Helmet Wearing Detection Algorithm Based on Improved YOLOX
Shuang Hao, Jianjun He, Li Luo, Kun Xi · 2022 5th International Conference on Pattern Recognition and Artificial Intelligence (PRAI) · 2022
Aiming at the importance of helmet detection in the current construction industry, an improved YOLOX helmet wearing detection algorithm is proposed. The MobileNetv3 network after optimizing the attention module is used to replace the backbone feature extraction network of YOLOX-s, so as to optimize the convergence ability of the model and greatly reduce the amount of network parameters; Then fusion factor in FPN is introduced to improve the detection accuracy of small targets. The experimental results on the helmet wearing dataset SHWD (safety helmet wearing dataset) show that the mAP value of FF-YOLO algorithm proposed in this paper reaches 91.84%, which is 1.21% higher than that of YOLOX-s model, and the parameters of this model are 37.8% lower than that of YOLOX-s model.