Insulator Fault Detection Using Improved Yolo V5

A power supply system’s stability is inextricably linked to the job of identifying problems in transmission lines. However, the insulator defect detection model based on deep learning is frequently employed in wire inspection. As a result, this study offers an enhanced YOLOv5s insulator defect detection model to address the issues of insufficient training data and low identification accuracy in the real-time detection of tiny target insulator flaws. To increase and improve the training data, tests were carried out using noise and random black blocks. The spatial and channel weight coefficients were generated by including an attention mechanism (Convolution Block Attention Module, CBAM), and the input feature map dimensions were altered to improve the model’s capacity to extract and fuse tiny target defect features. Experiments show that with Faster RCNN, YOLOv3, SSD, and YOLOv4 comparison experiments, the algorithm achieves 97.38% detection accuracy for insulators and 93.32% detection accuracy for small target insulator defects with a fast detection speed, which is a better solution to the problem of detecting insulator defects with an insufficient proportion in the image.