Abstract: During the operation of high-voltage transmission lines, defects may occur, and if not dealt with in time, the reliability of the line's power supply may be affected. Based on this, this paper proposes a defect detection method for high-voltage transmission lines based on the improved YOLOv8. First, the deformable convolution DCNv3 module is introduced to improve the traditional YOLOv8 algorithm to achieve nonlinear sampling of input features in the convolution process and improve the extraction ability of target features; secondly, the stable intersection ratio (SIoU) loss function is used for network optimization to improve the algorithm's defect detection level and accuracy; finally, the four typical defects of the actual transmission line defect data set are combined to verify and compared with other identification algorithms. The results show that the proposed improved algorithm has an accuracy rate of 95.82% for the detection of defects in high-voltage transmission lines, and the effectiveness of the proposed improved algorithm is verified.