基于YOLOv8的改进绝缘子缺陷检测算法  

Insulator Defect Detection Based on Improved YOLOv8

作　　者：宋文清 胡永祥[1] SONG Wen-qing;HU Yong-xiang(Hunan University of Technology,Zhuzhou 412007,China)

机构地区：[1]湖南工业大学,株洲412007

出　　处：《价值工程》2025年第6期104-107,共4页Value Engineering

摘　　要：针对现有绝缘子缺陷检测方法存在检测精度低,计算复杂度大等问题,提出了一种基于改进YOLOv8的绝缘子缺陷检测算法。首先,在骨干网络中,加入了C2f_SCConv模块,以增强对复杂背景中绝缘子缺陷的特征选择能力,有效减少背景噪声的干扰,从而提升整体的检测精度。其次,针对多尺度绝缘子缺陷检测的需求,在主干网络的关键部分嵌入了FocalModulation模块,该模块通过结合局部和全局特征调制,提升了模型在复杂场景下对小目标的检测能力。此外,为了满足轻量化设计和实时检测的要求,模型在骨干网络和检测头部分引入了Detect_Efficient模块,大幅减少了计算复杂度和参数量,从而提高了检测速度,同时保持了较高的检测精度。实验结果表明,改进后的算法相比原算法精度提高2.4%,召回率提高1.0%。通过与其他主流目标检测算法比较,验证了该方法的有效性。To address the issues of low detection accuracy and high computational complexity in existing insulator defect detection methods,an insulator defect detection algorithm based on improved YOLOv8 is proposed.Firstly,the C2f_SCConv module is added to the backbone network to enhance the feature selection capability for insulator defects in complex backgrounds,effectively reducing background noise interference and improving overall detection accuracy.Secondly,to meet the requirements of multi-scale insulator defect detection,the FocalModulation module is embedded in key parts of the backbone network.This module enhances the model's ability to detect small targets in complex scenes by modulating local and global features.Additionally,to satisfy the needs for lightweight design and real-time detection,the Detect_Efficient module is introduced in both the backbone network and detection head,significantly reducing computational complexity and the number of parameters,thereby increasing detection speed while maintaining high accuracy.Experimental results show that the improved algorithm achieves 2.4%increase in precision,1.0%increase in recall.The effectiveness of this method is validated through comparisons with other mainstream object detection algorithms.

关 键 词：目标检测 绝缘子缺陷 深度学习 YOLOv8 

分 类 号：TM216[一般工业技术—材料科学与工程]