Isolator Detection in Power Transmission Lines using Lightweight Dept-wise Convolution with BottleneckCSP YOLOv5

Abstract

The detection of insulators is of great importance in power transmission lines. This is because accurate detection ensures reliability and continuity of energy transmission, preventing line interruptions. The proposed method in this study utilizes the DWB-YOLOv5 (Dept-wise convolution with BottleneckCSP YOLOv5) model to effectively detect insulators, contributing to the safe and uninterrupted operation of power lines. In the suggested approach, the DWB-YOLOv5 model is employed to detect insulators. The bottleneckCSP module enhances the accuracy of targets at various scales, while the depth-wise c2onvolution module assists in reducing the model's complexity. Images undergo preprocessing steps such as automatic orientation and resizing. The preprocessed images are fed into the DWB-YOLOv5 model to extract deep features, perform object detection, and conduct classification. The insulator detection model obtained through this method exhibits a minimum of 8.53% better mean average precision (mAP) performance compared to existing methods. This study represents a significant step towards ensuring the safe and uninterrupted operation of power transmission lines. Accurate detection of insulators facilitates the smooth functioning of lines, ensuring reliability and continuity in energy transmission. The proposed method offers important advantages such as high accuracy, lightweight design, and efficiency.