Power Transmission Line Defect Recognition Method Based on Binocular Feature Fusion and Improved FCOS Detection Head

UAVs are widely used in transmission line inspection. Inspectors operate UAVs to take images of transmission lines and analyze and identify these images. Detecting transmission line defects in UAV inspection images is an important task. This paper proposes a transmission line defect detection method based on binocular feature fusion and an improved FCOS detection head. First, a binocular feature fusion module is designed. Second, a feature screening module is added to the network. Finally, add the IoU prediction branch to the FCOS detection head. The experimental results show that the transmission line defect detection method proposed in this paper can effectively identify the two defects of broken strands and foreign objects, and the mAP reaches 90.85%.

I. Introduction

Due to long-term exposure to the outside world, transmission lines sometimes have defects such as broken strands and hanging foreign objects. Identifying these defects in drone inspection images is crucial so that workers can deal with them in time.