Coal Shearer Drum Teeth Absence Detection Using Deep Learning Method

With the rapid development of the coal industry, the integrity and performance of coal shearer drum teeth are crucial for enhancing coal mining efficiency and ensuring operational safety. This paper proposes a method for detecting the absence of shearer drum teeth based on improved YOLOv8. First, the SimSPPF structure is integrated into the backbone feature extraction network of YOLOv8, where the SiLU activation function is replaced by ReLU in SimSPPF, significantly improving the efficiency of feature extraction. Furthermore, we incorporate a channel prior convolutional attention (CPCA) mechanism into the neck part of YOLOv8, which dynamically adjusts the importance of different channels using a channel attention mechanism, thereby enhancing the model's focus on crucial features. Then, to address the limitations of the CIoU loss function in boundary box regression, we adopt the WIoU loss function as a replacement. Experimental results show that compared with YOLOv8, the proposed method achieves a 2.5% increase in precision and a 2.7% improvement in mAP@0.5. Additionally, the proposed method achieves a detection speed of 135 frames per second, effectively enabling the detection of absence shearer drum teeth.