A Lightweight Insulator Defect Detection Model Based on YOLOv8

Aiming at the problems of lightweight insulator defect detection towards edge devices and insulator defect detection for small-scale defects, this paper proposes a lightweight insulator defect detection model that enhances YOLOv8. The model achieves the lightweight of the backbone network by incorporating ShuffleN etV2, thereby reducing the number of parameters and enhancing its suitability for lightweight operation on edge devices. Additionally, the introduction of the Cross-Scale Feature Fusion Module (CCFM) within the neck network enhances the model's capacity to detect small-scale insulator defects. Experimental results demonstrate the efficacy of the proposed model, with a recall rate of 97.3%, mAP@0.5reaching 99.1%, a parameter count of 2.7M, and a computational load of 8.2GFLOPs. Compared to the original model, the improved model is more suitable for deployment to edge devices for real-time detection of insulator defects.