Fig. 1. Network architecture of YOLOv8s

Fig. 2. Network structure of GER-YOLO

Fig. 3. Ghost module

Fig. 4. DFC long range attention module

Fig. 5. GhostNetV2 bottleneck module

Fig. 6. C2fGhostV2 module

Fig. 7. EMA network

Fig. 8. RFE module

Fig. 9. C2fRFE module

Fig. 10. Example of dataset

Fig. 11. Comparison of loss curves between GER-YOLO and YOLOv8s

Fig. 12. Visual comparison of detection results. (a) Detection effect of self-exploding insulators; (b) detection effect of damaged insulators; (c) detection effect of bird's nest; (d) detection effect of insulators with similar environmental colors