Improved YOLOv7-Tiny Insulator Defect Detection Based on Drone Images

Insulator defect detection is of great significance in power equipment maintenance and safety management. To address this problem, this paper proposes an insulator defect detection method based on the YOLOv7-tiny improved network. First, improvements have been made to the ELAN (Enhanced Lightweight Aggregation Network) module of YOLOv7-tiny. On the basis of the original ELAN module, the traditional convolution used for feature extraction was replaced by DWConv (depth separable convolution), forming the ELAN-DW module. This improvement not only effectively reduces the number of parameters and calculations of the model, but also increases the inference speed of the model. Secondly, this paper embeds a hybrid attention mechanism (CBAM) into ordinary convolutions, making it an attention convolution to replace the ordinary convolution at the neck of the network. Experimental results show that the improved method proposed in this paper has achieved significant performance improvement in the insulator defect detection task. In different scenarios, the model can accurately detect insulator defects with high accuracy and robustness. The improvement method proposed in this article provides an effective solution for actual power system maintenance and safety management, and has high application value and practicability.