计算机科学

计算机科学 ›› 2024, Vol. 51 ›› Issue (11A): 231200063-7.doi: 10.11896/jsjkx.231200063

• 图像处理&多媒体技术 • 上一篇    下一篇

基于YOLOv5的桥梁裂纹检测方法研究

李军1, 刘念2, 张世义2   

  1. 1 重庆交通大学机电与车辆工程学院 重庆 400074
    2 重庆交通大学航运与船舶工程学院 重庆 400074
  • 出版日期:2024-11-16 发布日期:2024-11-13
  • 通讯作者: 刘念 (2424092885@qq.com)
  • 作者简介:(qleejun@163.com)
  • 基金资助:
    国家自然科学基金(51305472);城市轨道交通车辆系统集成与控制重庆市重点实验室基金项目

Study on Detection Method of Bridge Crack Based on YOLOv5

LI Jun1, LIU Nian2, ZHANG Shiyi2   

  1. 1 School of Mechatronics and Vehicle Enginerring,Chongqing Jiaotong University,Chongqing 400074,China
    2 School of shipping and Naval Architecture,Chongqing JiaotongUniversity,Chongqing 400074,China
  • Online:2024-11-16 Published:2024-11-13
  • About author:LI Jun,born in 1964,Ph. D，professor, master's supervisor.His main research interests include energy saving and new energy vehicles and their applications,intelligent vehicle perception，decision making and control, automobile engine combustion emissions and control, advanced manufacturing technology and application.
    LIU Nian,born in 1999,postgraduate.His main research interests include bridge damage crack identification and detection,and so on.
  • Supported by:
    National Natural Science Foundation of China (51305472) and Chongqing Key Laboratory of Urban Rail Transit Vehicle System Integration and Control Fund Project.

PDF (PC)

170

摘要： 为解决桥梁裂纹识别中不同裂纹的识别问题,提高模型的拟合能力,并提升裂纹的特征提取能力,提出了一种基于YOLOV5融合EfficientNet,引入CBAM注意力机制的桥梁裂纹识别的算法YOLOv5-Crack。首先,基于网络EfficientNet的高精度、高效性,替换YOLOv5的特征提取网络以提取裂纹特征;其次,将CBAM(Convolutional Block Attention Module)卷积块注意力模块与通道和空间注意力模块结合,以增强模型表达浅层目标特征信息的能力,提高了裂纹的识别精度;最后,在桥梁裂缝数据集Concrete Crack Images for Classification上训练。研究结果表明:在大型裂纹的识别能力上,YOLOv5-Crack检测识别精度高于YOLOv5,其mAP@0.5,Recall及Precision明显提高,而消耗的算力明显降低,能够满足裂纹的检测要求。

关键词: YOLOV5, EfficientNet, 裂纹识别, CBAM

Abstract: To address the issues of different crack recognition in bridge crack identification,improve the model's fitting ability,and enhance crack feature extraction capability,this paper proposes an algorithm called YOLOv5-Crack based on the fusion of YOLOv5 and EfficientNet,incorporating the CBAM attention mechanism in bridge crack recognition.Firstly,the feature extraction network of YOLOv5 is replaced with the EfficientNet network known for its high accuracy and efficiency,to extract crack features.Secondly,the convolutional block attention module(CBAM) is used to enhance the model's ability to capture the feature information of shallow targets by combining channel and spatial attention modules,thereby improving crack recognition accuracy.Finally,the model is trained on the bridge crack dataset “concrete crack images for classification”.The research results show that YOLOv5-Crack demonstrates higher accuracy in detecting large cracks compared to YOLOv5,with improved mAP@0.5,recall,and precision.Additionally,it consumes less computing power while meeting the requirements of crack detection.

Key words: YOLOv5, EfficientNet, Crack detection, CBAM

中图分类号: 

  • U448.33
[1]WANG X Q.Fatigue crack detection and evaluation of steel structure bridges based on AC electromagnetic field detection technology[J].Industrial Architecture,2023,53(8):102-106.
[2]LIU X,GAO S W,HE Y.Bridge Crack Detection and Recognition Based on Convolutional Neural Network Transfer Learning [J].Science and Technology Innovation Herald,2019(4):24-25.
[3]SHAO Z X.Research and Implementation of Concrete Ultrasonic Testing Technology [J].Vibration,Testing and Diagnosis,2012,32(3):397-401,513.
[4]DENG Z Y,KANG Y H,ZHANG J K,et al.Multi-source effect in magnetizing-based eddy current testing sensor for surface crack in ferromagnetic materials[J].Sensors & Actuators:A.Physical,2018,271.
[5]QUANG T L,NAOYA K,KOUICHI S,et al.Eddy current convergence probes with self-differential and self-nulling characteristics for detecting cracks in conductive materials[J].Sensors and Actuators:A.Physical,2023,349.
[6]ZHANG X Y,ZHOU B,LI H,et al.Depth detection of spar cap defects in large-scale wind turbine blades based on a 3D heat conduction model using step heating infrared thermography[J].Measurement Science and Technology,2022,33(5).
[7]DU Y C,ZHANG X M,LI F,et al.Detection of Crack Growth in Asphalt Pavement Through Use of Infrared Imaging[J].Transportation Research Record Journal of the Transportation Research Board,2017,2645:24-31.
[8]TANG L,JIANG J P,GU P Y,et al.Experimental study on ultrasonic infrared detection of concrete components[J].Journal of Hydraulic Engineering,2012,43(Z1):70-75.
[9]XU Y,ZHANG H Y,WANG Q Y.Experimental study on microcrack detection of heat-damaged concrete based on nonlinear ultrasonic technology[J].Journal of Vibration and Shock,2021,40(5):126-135.
[10]NITHIN K,RAVIRAJ V,RENGASWAM Y J.Machine Learning Based Predictions of Fatigue Crack Growth Rate of Additively Manufactured Ti6Al4V[J].Metals,2021,12(1).
[11]ZHANG L,WANG Z C,WANG L,et al.Machine learning-based real-time visible fatigue crack growth detection[J].Digital Communications and Networks,2021,7(4).
[12]MIR B A,SASAKI T,NAKAO K,et al.Machine learning-based evaluation of the damage caused by cracks on concrete structures[J].Precision Engineering,2022,76.
[13]WANG C L,HOU X L,LIU Y B.Three-Dimensional CrackRecognition by Unsupervised Machine Learning[J].Rock Mechanics and Rock Engineering,2021:893-903.
[14]LI H F,SONG D Z,LIUY,et al.Automatic Pavement Crack Detection by Multi-Scale Image Fusion[J].IEEE Trans.Intelligent Transportation Systems,2019,20(6).
[15]ZHANG H C,QIAN Z D,TAN Y F,et al.Investigation of pavement crack detection based on deep learning method using weakly supervised instance segmentation framework[J].Construction and Building Materials,2022,358.
[16]ZHAO C Y,ZHANG H,LIAO D,et al.Rail surface defect detection model based on attention mechanism and mixed supervised learning[J].Computer Science,2022,49(S2):488-493.
[17]TENG S,LIU Z,CHEN G,et al.Concrete Crack Detection Basedon Well-Known Feature Extractor Model and the YOLO_v2 Network[J].Applied Sciences,2021,11(2):813.
[18] DU Y C,PAN N,XU Z H,et al.Pavement distress detection and classification based on YOLO network[J].International Journal of Pavement Engineering,2020(1):1-14.
[19]ZHU X,ZHU M,REN H.Method of plant leaf recognitionbased on improved deep convolutional neural network[J].Cognitive Systems Research,2018,52(DEC.):223-233.
[20]WANG M,WANG K,LI S,et al.Pill detection algorithm based on improved EfficienDet[J].Electronic Measurement Technology,2022,45(19):136-142.
[21]REDMON J,DIVVALA S,GIRSHICKR,et al.You Only Look Once:Unified,Real-Time Object Detection[C]// Computer Vision & Pattern Recognition.IEEE,2016.
[22]TAN M X,LE Q V.EfficientNet:Rethinking Model Scaling for Convolutional Neural Networks[J].arXiv:1905.11946,2019.
[23]ZHANG S,LIU Z,CHEN Y,et al.Selective kernel convolution deep residual network based on channel-spatial attention mechanism and feature fusion for mechanical fault diagnosis[C]//ISA Transactions.2022.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市渝北区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发