Research on Illegal Behavior Detection Algorithm of Underground Mine Workers Based on Improved YOLOv8

Aiming at the problems of poor detection effect and weak generalization ability of illegal production of workers in mines, a mine worker illegal target detection algorithm based on improved YOLOv8 is proposed. In view of the problem of large scale changes of targets in mines, a unified self-attention mechanism detection head (DyHead) is introduced to replace the original detection head, which greatly improves the feature expression ability of the detection head. Experimental results show that on the public dataset VOC2012, after adding DyHead to yolov8, the detection accuracy is improved by 1.5%, on the self-built dataset, after adding DyHead, the detection accuracy is improved by 2.3 %. This shows that compared with the original model, the improved algorithm after adding DyHead not only greatly improves the model detection ability, but also has good generalization performance. Compared with the mainstream algorithm, the algorithm proposed in this paper has lower model complexity and higher detection accuracy, which is suitable for target detection in complex environments in mines.