A deep learning-based algorithm for surface crack detection of concrete bridge

doi:10.3969/j.issn.1000-1565.2026.02.009

Abstract

Abstract: Aiming at the issues of low recognition accuracy and slow detection speed in traditional bridge crack detection methods, a deep learning-based algorithm for surface crack detection of concrete bridge is proposed in this paper. In this algorithm, a feature encoder based on Ghost convolution is designed, which greatly reduces the parametric number of the network while improving the identification accuracy; secondly, a lightweight multi-scale feature extraction module based on SimAM is proposed, which- 引用格式:何梦腾,潘浩文,邓洪兴,等.基于点云多尺度方向一致性的羊只体尺测量方法［J］.河北大学学报(自然科学版),2026,46(2):113-127.引用格式:袁正学,陈琛,林昆朋,等.基于深度学习的混凝土桥梁表面裂缝识别算法［J］.河北大学学报(自然科学版),2026,46(2):204-214.DOI:10.3969/j.issn.1000-1565.2026.02.009基于深度学习的混凝土桥梁表面裂缝识别算法袁正学¹,陈琛²,林昆朋¹,许宝峰¹,郭一鹏³(1. 中铁贵州工程有限公司,贵州贵安 561116;2. 湘潭大学土木工程学院,湖南湘潭 411105;3. 长沙理工大学土木与环境工程学院,湖南长沙 410114)摘要:针对传统桥梁裂缝检测方法中存在的识别精度低、检测速度慢等问题,提出了一种基于深度学习的混凝土桥梁表面裂缝识别算法.在该算法中,首先设计了一种基于Ghost卷积的特征编码器,在提升裂缝识别精度的同时,大大减小了网络的参数量;其次,提出了一种基于SimAM增强的轻量化多尺度特征提取模块,有效减少裂缝识别过程中复杂背景干扰(蜂窝、麻面、噪音、手写标记)造成的错检和漏检问题,并提升了网络对于不同尺度裂缝的特征提取能力;最后,采用参数可学习的DUpsampling代替特征解码模块中的传统线性插值上采样操作,以输出更加精确的像素预测结果.实验结果表明,本文提出桥梁裂缝识别算法的精度指标mPA和mIoU分别为81.54%和86.77%,相较于DeepLabv3+、Unet、Segformer、Swin-Unet等4种常用裂缝识别算法具有明显的提升.此外,本文模型的图像处理速度FPS为43.91 f/s,模型大小仅为46.9 MB,很好地满足移动设备实时检测的指标要求,适用于桥梁裂缝的智能化、高精度、快速识别.关键词:桥梁工程;裂缝识别;语义分割;深度学习中图分类号:U455 文献标志码:A 文章编号:1000-1565(2026)02-0204-11DOI:10.3969/j.issn.1000-1565.2026.02.009A deep learning-based algorithm for surface crack detection of concrete bridgeYUAN Zhengxue¹, CHEN Chen², LIN Kunpeng¹, XU Baofeng¹, GUO Yipeng³(1. China Railway Guizhou Engineering Co., Ltd., Guian 561116, China; 2. College of Civil Engineering, Xiangtan University, Xiangtan 411105,China;3. School of Civil Environmental Engineering, Changsha University of Science & Technology, Changsha 410114,China)Abstract: Aiming at the issues of low recognition accuracy and slow detection speed in traditional bridge crack detection methods, a deep learning-based algorithm for surface crack detection of concrete bridge is proposed in this paper. In this algorithm, a feature encoder based on Ghost convolution is designed, which greatly reduces the parametric number of the network while improving the identification accuracy; secondly, a lightweight multi-scale feature extraction module based on SimAM is proposed, which- 收稿日期:2024-12-16;修回日期:2025-11-14 基金项目:国家自然科学基金青年项目(52008038);湖南省自然科学基金青年项目(2024JJ6429);中铁二十局集团公司科技研发重大专项(YF2513LJ03A);中铁贵州工程有限公司科技研发项目(YF2313QL01D) 第一作者:袁正学(1976—),男,中铁贵州工程有限公司高级工程师,主要从事铁路、公路等施工技术方向研究.E-mail:zhengx0111@163.com 通信作者:陈琛(1989—),男,湘潭大学讲师,博士,主要从事铁路隧道工程、基础工程等方向研究.E-mail:chenchen19892024@126.com 第2期袁正学等:基于深度学习的混凝土桥梁表面裂缝识别算法河北大学学报(自然科学版) 第46卷effectively decrease the misdetection and omission problems caused by the complex background disturbances(honeycomb, sisal, noise, and handwritten markings), and improves the networks feature extraction capability for cracks at different scales. Finally, parameter-learning DUpsampling is used to replace the traditional linear interpolation upsampling operation in the feature decoding module to output more accurate pixel prediction results. The experimental results show that the accuracy metrics mPA and mIoU of the proposed algorithm in this paper are 81.54% and 86.77%, respectively, which are significantly improved compared with DeepLabv3+, Unet, Segformer and Swin-Unet. In addition, FPS of the model in this paper is 43.91 f/s, and the model size is only 46.9 MB, which well meets the index requirements of real-time detection of mobile devices, and is suitable for the intelligent, high-precision, and fast identification of bridge cracks.

Key words: bridge engineering, crack identification, semantic segmentation, deep learning

CLC Number:

U455

YUAN Zhengxue, CHEN Chen, LIN Kunpeng, XU Baofeng, GUO Yipeng. A deep learning-based algorithm for surface crack detection of concrete bridge[J]. Journal of Hebei University(Natural Science Edition), 2026, 46(2): 204-214.

References

[1] 何昊,贺福强,谢丹,等.基于SVM和改进区域生长法的桥梁裂缝分割算法[J].公路交通科技, 2022, 39(11): 115-123. DOI:10.3969/j.issn.1002-0268.2022.11.015.
[2] Prasanna P, Dana K J, Gucunski N, et al. Automated crack detection on concrete bridges[J]. IEEE Trans Autom Sci Eng, 2016, 13(2): 591-599. DOI:10.1109/TASE.2014.2354314.
[3] Vivekananthan V, Vignesh R, Vasanthaseelan S, et al. Concrete bridge crack detection by image processing technique by using the improved OTSU method[J]. Mater Today Proc, 2023, 74: 1002-1007. DOI:10.1016/j.matpr.2022.11.356.
[4] Xu H Y, Su X, Wang Y, et al. Automatic bridge crack detection using a convolutional neural network[J]. Appl Sci, 2019, 9(14): 2867. DOI:10.3390/app9142867.
[5] 朱苏雅,杜建超,李云松,等.采用U-Net卷积网络的桥梁裂缝检测方法[J].西安电子科技大学学报, 2019, 46(4): 35-42. DOI:10.19665/j.issn1001-2400.2019.04.006.
[6] 余加勇,李锋,薛现凯,等.基于无人机及Mask R-CNN的桥梁结构裂缝智能识别[J].中国公路学报, 2021, 34(12): 80-90. DOI:10.19721/j.cnki.1001-7372.2021.12.007.
[7] Zhang J, Qian S R, Tan C. Automated bridge surface crack detection and segmentation using computer vision-based deep learning model[J]. Eng Appl Artif Intell, 2022, 115: 105225. DOI:10.1016/j.engappai.2022.105225.
[8] 杨燕泽,王萌,刘诚,等.基于语义分割的沥青路面裂缝智能识别[J].浙江大学学报(工学版), 2023, 57(10): 2094-2105. DOI:10.3785/j.issn.1008-973X.2023.10.018.
[9] Zhou Z, Zhang J J, Gong C J. Automatic detection method of tunnel lining multi-defects via an enhanced You Only Look Once network[J]. Computer Aided Civil Eng, 2022, 37(6): 762-780. DOI:10.1111/mice.12836.
[10] Su H F, Wang X, Han T, et al. Research on a U-Net bridge crack identification and feature-calculation methods based on a CBAM attention mechanism[J]. Buildings, 2022, 12(10): 1561. DOI:10.3390/buildings12101561.
[11] Asadi Shamsabadi E, Xu C, Rao A S, et al. Vision transformer-based autonomous crack detection on asphalt and concrete surfaces[J]. Autom Constr, 2022, 140: 104316. DOI:10.1016/j.autcon.2022.104316.
[12] Han K, Wang Y H, Tian Q, et al. GhostNet: more features from cheap operations[C] //2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR). June 13-19, 2020, Seattle, WA, USA. IEEE, 2020: 1577-1586. DOI:10.1109/CVPR42600.2020.00165.
[13] Yang L, Zhang R Y, LI L, et al. Simam: A simple, parameter-free attention module for convolutional neural networks[C] //International conference on machine learning. PMLR, 2021: 11863-11874.
[14] Tian Z, He T, Shen C H, et al. Decoders matter for semantic segmentation: data-dependent decoding enables flexible feature aggregation[C] //2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR). June 15-20, 2019. Long Beach, CA, USA. IEEE, 2019: 3126-3135. DOI:10.1109/cvpr.2019.00324.
[15] Liu R R, Tao F, Liu X T, et al. RAANet: a residual ASPP with attention framework for semantic segmentation of high-resolution remote sensing images[J]. Remote Sens, 2022, 14(13): 3109. DOI:10.3390/rs14133109.
[16] Chu H H, Wang W, Deng L. Tiny-crack-net: a multiscale feature fusion network with attention mechanisms for segmentation of tiny cracks[J]. Computer Aided Civil Eng, 2022, 37(14): 1914-1931. DOI:10.1111/mice.12881.
[17] Zhou Z, Zhang J J, Gong C J. Hybrid semantic segmentation for tunnel lining cracks based on Swin Transformer and convolutional neural network[J]. Computer Aided Civil Eng, 2023, 38(17): 2491-2510. DOI:10.1111/mice.13003.
[19] Chen L C, Zhu Y K, Papandreou G, et al. Encoder-decoder with atrous separable convolution for semantic image segmentation[M] //Computer Vision-ECCV 2018. Cham: Springer International Publishing, 2018: 833-851. DOI:10.1007/978-3-030-01234-2_49.
[20] Zhao H S, Shi J P, Qi X J, et al. Pyramid scene parsing network[C] //2017 IEEE Conference on Computer Vision and Pattern Recognition(CVPR). July 21-26, 2017, Honolulu, HI, USA. IEEE, 2017: 6230-6239. DOI:10.1109/CVPR.2017.660.
[21] Xie E, Wang W, Yu Z, et al. SegFormer: Simple and efficient design for semantic segmentation with transformers[J]. Advances in Neural Information Processing Systems, 2021, 34: 12077-12090.
[22] Cao H, Wang Y Y, Chen J, et al. Swin-unet: unet-like pure transformer for medical image segmentation[M] //Computer Vision-ECCV 2022 Workshops. Cham: Springer Nature Switzerland, 2023: 205-218. DOI:10.1007/978-3-031-25066-8_9.
[23] Hu J, Shen L, Sun G. Squeeze-and-excitation networks[C] //2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. June 18-23, 2018, Salt Lake City, UT, USA. IEEE, 2018: 7132-7141. DOI:10.1109/CVPR.2018.00745.
[24] Zhu X Z, Cheng D Z, Zhang Z, et al. An empirical study of spatial attention mechanisms in deep networks[C] //2019 IEEE/CVF International Conference on Computer Vision(ICCV). October 27-November 2, 2019. Seoul, Korea(South). IEEE, 2019: 6688-6697. DOI:10.1109/iccv.2019.00679.
[25] Woo S, Park J, Lee J Y, et al. CBAM: convolutional block attention module[C] // Computer Vision-ECCV 2018. Cham: Springer International Publishing, 2018: 3-19. DOI:10.1007/978-3-030-01234-2_1. (