南方电网技术2026，Vol.20Issue(4)：140-152,13.DOI:10.13648/j.cnki.issn1674-0629.2026.04.013
基于改进YOLOv8n的输电线路绝缘子缺陷检测方法
Defect Detection Method for Transmission Line Insulator Based on Improved YOLOv8n
摘要
Abstract
Due to the complex backgrounds,small defect targets,and significant size variations among different detection objects of insulator defects in drone aerial inspection images,current object detection algorithms often lose defect features when identifying such targets,resulting in low accuracy and high misidentification rates.To address this,an improved YOLOv8n-based method for detect-ing insulator defects in transmission lines is proposed.Firstly,to tackle the feature loss issue of small insulator defects,a PCFocalNeXt-C2f module is designed and used to optimize the Backbone of YOLOv8n.Secondly,to mitigate feature information loss or degradation during multi-level transmission,an asymptotic feature pyramid network(AFPN)supporting direct interaction across non-adjacent layers is employed to enhance the Neck of YOLOv8n.Finally,the Sigmoid-CIoU loss function is introduced to optimize the loss function in YOLOv8n,and Sigmoid-CIoU NMS is adopted to obtain detection results,improving the model's robust-ness.Experimental results show that compared to the original YOLOv8n,the proposed algorithm achieves a 3.3%increase in F1-score,with mAP@0.5 and mAP@0.5-0.95 improving by 3.8%and 6.1%,respectively,while reducing the parameter count by 23.9％,demonstrating superior detection performance.关键词
无人机巡检/缺陷检测/YOLOv8n/小目标检测/AFPNKey words
unmanned aerial vehicle inspection/defect detection/YOLOv8n/small target detection/AFPN分类
信息技术与安全科学引用本文复制引用
王爽,李亚威,阴酉龙,唐波..基于改进YOLOv8n的输电线路绝缘子缺陷检测方法[J].南方电网技术,2026,20(4):140-152,13.基金项目
国家自然科学基金资助项目(72271076) （72271076）
湖北省重点研发项目(2020BAB110). Supported by the National Natural Science Foundation of China(72271076) （2020BAB110）
the Key Research and Development Projects of Hubei Province(2020BAB110). （2020BAB110）
