基于ECA和YOLOv5的隧道渗水检测方法  

Tunnel water leakage detection method based on ECA and YOLOv5

作　　者：杨丽[1,2] 邓靖威 段海龙[1,2] 杨晨晨 YANG Li;DENG Jingwei;DUAN Hailong;YANG Chenchen(School of Automation and Electrical Engineering,Tianjin University of Technology and Education,Tianjin 300222,China;Tianjin Key Laboratory of Information Sensing and Intelligent Control,Tianjin University of Technology and Education,Tianjin 300222,China)

机构地区：[1]天津职业技术师范大学自动化与电气工程学院,天津300222 [2]天津职业技术师范大学天津市信息传感与智能控制重点实验室,天津300222

出　　处：《天津职业技术师范大学学报》2024年第2期19-24,共6页Journal of Tianjin University of Technology and Education

基　　金：天津市教委科研计划项目(2022ZD036);天津市自然科学基金资助项目(20JCZDJC00150)。

摘　　要：针对现有方法存在的隧道渗水检测精度不高和特征融合过程中信息丢失的问题,提出了一种基于有效的通道注意力(efficient channel attention,ECA)和YOLOv5的隧道渗水检测方法。该方法融合ECA注意力模块设计瓶颈结构,加强挖掘浅层特征表征的几何结构信息,充分提取水迹特征信息的同时抑制背景特征,提高水迹检测精度。在建立的隧道渗水水迹数据集上进行实验,结果表明:对比原YOLOv5模型,所提出的隧道渗水检测方法的平均精度均值提高了10%,准确率提高了17%,召回率提高了6%。实验结果验证了该方法的有效性。To solve the problems of low detection accuracy and information loss in feature fusion process for methods of the existing tunnel water seepage detection,a tunnel water seepage detection method based on ECA(efficient channel attention,ECA)and YOLOv5 is proposed,which fuses the ECA attention module to design the bottleneck structure,enhances the mining of geometric structure information with shallow feature representation,fully extracts water stain feature information while suppressing background features,and improves water stain detection accuracy.Experimental results using an established water trace data set of tunnel seepage demonstrate that the proposed method outperforms the original YOLOv5 model.Specifically,the proposed tunnel seepage detection method achieves a 10%increase in average accuracy,a 17%increase in accuracy rate,and a 6%increase in recall rate,validating its effectiveness in tunnel seepage detection.

关 键 词：隧道渗水检测 深度学习 注意力机制 

分 类 号：U456[建筑科学—桥梁与隧道工程]