[1]潘学华,梁炜皓,李金瑾,等.面向绝缘子微小缺陷的 YOLOv8s 的改进算法研究[J].机械与电子,2025,(12):10-17.
　PAN Xuehua,LIANG Weihao,LI Jinjin,et al.Research on Improved YOLOv8s Algorithm for Micro-defect Detection in Insulators[J].Machinery & Electronics,2025,(12):10-17.
点击复制

面向绝缘子微小缺陷的 YOLOv8s 的改进算法研究()
分享到：

《机械与电子》[ISSN:1001-2257/CN:52-1052/TH]

卷:
期数:
2025年12期
页码:
10-17
栏目:
研究与设计
出版日期:
2025-12-23

文章信息/Info

Title:
Research on Improved YOLOv8s Algorithm for Micro-defect Detection in Insulators
文章编号:
1001-2257 ( 2025 ) 12-0010-08
作者:
潘学华 1 ; 梁炜皓 1 ; 李金瑾 1 ; 张 震 1 ; 龚宇平 1 ; 尹显贵 2 ; 刘益畅 2
1. 广西电网有限责任公司计量中心,广西 南宁 530000 ;
2. 浪潮通信信息系统有限公司,山东 济南 250014
Author(s):
PAN Xuehua1 ;  LIANG Weihao1 ;  LI Jinjin1 ;  ZHANG Zhen1 ;  GONG Yuping1 ;  YIN Xiangui2 ;  LIU Yichang2
( 1.Measurement Center , Guangxi Power Grid Co. , Ltd. , Nanning 530000 , China ;?
2.Inspur Communication Information System Co. , Ltd. , Jinan 250014 , China
关键词:
绝缘子缺陷;  YOLOv8s ; 注意力机制; 小目标检测
Keywords:
insulator defects ;  YOLOv8s ;  attention mechanism ;  small object detection
分类号:
TP391.4
文献标志码:
A
摘要:
针对绝缘子表面缺陷检测中存在的小尺度目标特征微弱、复杂背景干扰等技术难题,提出一种基于 YOLOv8s 架构的轻量化与细粒度特征增强相结合的检测框架 YOLOv8-CEDA 。通过设计 GCA 模块,融合 Ghost 模块的轻量化优势与 Coordinate Attention 的空间信息建模能力,在显著降低模型参数和计算复杂度的同时,有效增强对微小缺陷的空间特征表征能力;在检测头中部署轻量级 ECA 通道注意力模块,实现关键特征通道的自适应权重分配,从而提升微弱缺陷信号的响应强度;此外,采用 DySample 动态上采样模块替代传统上采样方法,使多尺度特征融合过程具备内容感知能力,进一步保持细节特征并增强边缘信息。实验结果表明,所提方法在保持模型轻量化特性的基础上,检测精度较基线模型显著提升,尤其在小目标缺陷检测任务中表现优异,展现出良好的工程应用价值。
Abstract:
Detecting small scale insulator surface defects remains challenging due to weak feature representation and complex background interference.To address this issue , this study proposes YOLOv8 CEDA , an enhanced lightweight object detection framework based on YOLOv8s , integrating fine grained feature enhancement.First , a Ghost and Coordinate Attention ( GCA ) module is designed , combining the parameter efficiency of Ghost convolution with the spatial modeling capability of Coordinate Attention to improve small defect localization while reducing computational overhead.Second , an efficient channel attention ( ECA ) mechanism is incorporated into the detection head to adaptively amplify discriminative features , strengthening the response to subtle defect patterns.Additionally , DySample , a dynamic upsampling operator , replaces conventional interpolation to preserve structural details and enhance edge recovery during multi-scale feature fusion.Extensive experiments demonstrate that YOLOv8-CEDA achieves superior detection accuracy over baseline models while maintaining computational efficiency , particularly in small defect detection.The proposed method exhibits strong potential for real-world industrial applications.

参考文献/References:

[ 1 ] 冷月香,王健 . 基于改进的 YOLOv5 绝缘子检测与识别算法[ J ] . 计算机测量与控制, 2025 , 33 ( 4 ): 95-101.

[ 2 ] FANG G Z , AN X , FANG Q , et al.Insulator defect detection based on YOLOv5s-KE [ J ] .Electronics ,2024 , 13 ( 17 ): 3483.
[ 3 ] LIU Y , HUANG X B.Efficient cross-modality insulator augmentation for multi-domain insulator defect detection in UAV images [ J ] .Sensors , 2024 , 24 ( 2 ): 428.
[ 4 ] ZHANG Z Y , HUANG S H , LI Y X , et al.Image detection of insulator defects based on morphological processing and deep learning [ J ] .Energies , 2022 , 15( 7 ): 2465.
[ 5 ] CHEN H Y , HU Q D , ZHAI B S , et al.A robust weakly supervised learning of deep Conv-Nets for surface defect inspection [ J ] .Neural computing and applications , 2020 , 32 ( 15 ): 11229-11244.
[ 6 ] 胡昭华,王长富 . 改进 Faster R-CNN 的遥感图像小目标检测算法[ J ] . 计算机工程与科学, 2024 , 46 ( 6 ): 1063-1071.
[ 7 ] 伍箴燎,吴正平,孙水发 . 基于改进 YOLOv5 算法的绝缘子多缺陷检测 [ J ] . 高压电器,2024 , 60 ( 12 ): 95-102 , 112.
?[ 8 ] 高英铭,韩顺杰,周波,等 . 改进 YOLOv8 的绝缘子及其自爆缺陷目标检测算法[ J ] . 无线电工程,2025 , 55( 2 ): 281-290.
[ 9 ] 廖丽瑛,刘洪 . 基于改进 YOLOv8 的绝缘子自爆缺陷检测[ J ] . 电子测量技术, 2024 , 47 ( 18 ): 138-144.
[ 10 ] 杨茜,熊炜,孟圣哲,等 . 基于改进 YOLOv8 的绝缘子缺陷检测方法[ J ] . 电子测量技术, 2025 , 48 ( 7 ): 86-97.
[ 11 ] 朱泓宇,程换新,骆晓玲 . 改进 YOLOv8 网络在绝缘子缺陷检测中的应用 [ J ] . 水电能源科学,2024 , 42( 5 ): 183-187.
[ 12 ] 王庆磊,杨宇航,陈立业,等 . 基于改进 YOLOv8n 算法的井下人员行为检测研究[ J ] . 煤矿机械, 2025 , 46( 5 ): 195-200.
[ 13 ] 柴威,杨明浩,韩沈飞,等 . 基于改进 YOLOv8 的监控中异常行为检测算法研究 [ J ] . 刑事技术, 2025 , 50( 2 ): 124-131.
[ 14 ] 姜源,付波,权轶,等 . 基于改进 YOLOv8n 的 PCB 缺陷检测算法[ J ] . 国外电子测量技术, 2024 , 43 ( 6 ): 22-32.
[ 15 ] 李文辰,张亚慧 . 基于 YOLOv5s 改进的机油瓶表面缺陷检测方法[ J ] . 计算机测量与控制, 2025 , 33 ( 4 ): 32-39.
[ 16 ] WANG J X , LU Y L , GUO J M , et al.SODD-YOLOv8 : an insulator defect detection algorithm based on feature enhancement and variable row convolution [ J ] . Measurement science and technology , 2024 , 36 ( 1 ): 015401.
[ 17 ] LIANG F , ZHAO L , REN Y , et al.LAD-Net : a lightweight welding defect surface non-destructive detection algorithm based on the attention mechanism [ J ] . Computers in industry , 2024 , 161 : 104109.
[ 18 ] LIU W Z , LU H , FU H T , et al.Learning to upsample by learning to sample [ C ] ∥2023 IEEE / CVF International Conference on Computer Vision ( ICCV ),2023 : 6027-6037.
[ 19 ] GARG K , NAYAR S K.Photorealistic rendering of rain streaks [ C ] ∥Proceedings of the European Conference on Computer Vision , 2007 : 184-197.
[ 20 ] PEARSON J K , PIEPER M.Real-time particle based snow simulation [ C ] ∥Proceedings of the 2012 ACM SIGGRAPH Symposium on Interactive 3D Graphics and Games , 2012 : 125-132.

备注/Memo

备注/Memo:
收稿日期: 2025-07-25
基金项目:广西电网公司科技项目资助( GXKJXM20230003 )
作者简介:潘学华 ( 1980- ),女,四川绵阳人,硕士,高级工程师,研究方向为电能计量;李金瑾 ( 1987- ),男,广西南宁人,硕士,高级工程师,研究方向为电力工业,通信作者, E-mail : tony3289@163.com 。
更新日期/Last Update: 2026-01-04