 The proposed algorithm is a shearer drum identification method based on improved YOLOv5s with dark channel guided filtering defogging. This algorithm uses the coordinate attention, CA, mechanism to improve the backbone network of the YOLOv5s algorithm. Additionally, the C3 module is used to extract shearer drum features from the image. These features are then reallocated by the attention mechanism to the weights of each space and channel. This allows for better information propagation of shearer drum features, resulting in improved target detection. Furthermore, the improved algorithm is compared against other target detection algorithms. The results show that the improved algorithm outperforms most target detection algorithms in terms of both accuracy and speed. This article was authored by Ching Huamao, Meng Han-Wang, XI and Hu, and others.