Abnormal cable detection at the bottom of High-speed Train Using Improved Yolov8

Cables are of great significance for high-speed trains, which are directly related to the operation safety, efficiency and reliability of high-speed trains. However, due to the complex train operation environment, cable defects occur from time to time, which may cause traffic accidents and endanger the lives of passengers. Therefore, it is very important to test the cables on high-speed trains. In this paper, YOLOv8 is improved. Firstly, the SPPF module in YOLOv8 Backbone is replaced by Focal Modulation module. Then, the Semantics and Detail Infusion Module (SDI) in U-Net v2 is used to redesign the feature fusion part in YOLOv8's Neck. Finally, the separation enhancement attention module in YOLO-FaceV2 is used to improve the Head of YOLOv8, so that it can effectively deal with occlusion scenes. Our method finally achieves 74.9% AP, which effectively improves the detection accuracy.