摘要
针对管道内壁缺陷检测精度低、易漏检等问题,提出一种基于改进YOLOv5的天然气管道内壁缺陷检测算法。首先,在主干网络与加强特征提取网络之间加入CBAM注意力机制,使网络关注到更加有效的信息,从而提高网络的特征提取能力;其次,将损失函数由GIOU替换为CIOU,以提升算法的精度和收敛速度。选取东北大学钢材表面缺陷数据集进行实验验证,结果表明,相较于YOLOv5算法,该算法的检测效果更好。
To address the problem of the low accuracy and easy missed detection in detecting defects of natural gas pipeline inner walls,an algorithm for defect detection of natural gas pipeline inner wall based on improved YOLOv5 is proposed.Firstly,a CBAM attention mechanism is added between the backbone network and the enhanced feature extraction network to enable the network to focus on more effective information,thereby improving the network′s feature extraction capability.Secondly,the loss function GIOU is replaced by CIOU in order to improve the accuracy and convergence speed of the algorithm.Then the data set of steel surface defects from Northeastern University was selected for experimental verification,and the results show that the proposed algorithm has better detection performance compared with the YOLOv5 algorithm.
作者
梁书溢
何亚平
唐德东
崔文岩
何小宇
周德
LIANG Shuyi;HE Yaping;TANG Dedong;CUI Wenyan;HE Xiaoyu;ZHOU De(School of Electrical Engineering,Chongqing University of Science and Technology,Chongqing 401331,China)
出处
《重庆科技学院学报（自然科学版）》
CAS
2023年第4期74-79,共6页
Journal of Chongqing University of Science and Technology：Natural Sciences Edition
基金
重庆市技术创新与应用发展专项重点项目“免维护天然气在线分析技术与应用推广”(CSTC2019JSCX-MSXMX0054)
重庆科技学院研究生创新项目“基于改进YOLOv7的带钢缺陷检测系统设计及其实现”(YKJCX2220408),“基于YOLOv7的智能交通监测系统”(YKJCX2220419)。
