A Review on Deep Learning-Based Segmentation Methods for Concrete Cracks in Bridges and Pavements

Chunlei Zhao; Ziqin Ye; Zhan Zhang; Yuanyuan Cao; Liwei Wu

doi:10.54691/f2fd3702

Authors

Chunlei Zhao
Ziqin Ye
Zhan Zhang
Yuanyuan Cao
Liwei Wu

DOI:

https://doi.org/10.54691/f2fd3702

Keywords:

Deep Learning, Segmentation, Multi-scale Feature Fusion, Attention Mechanism.

Abstract

As critical transportation infrastructure, the structural integrity of bridges and pavements directly impacts the safety and service life of transportation systems. Traditional manual crack detection methods suffer from inefficiency, subjectivity, and difficulties in result quantification. In recent years, the rapid advancement of computer vision technologies and artificial intelligence theories has provided effective solutions to these challenges, particularly through intelligent crack segmentation methods based on deep learning, which have demonstrated significant advantages.This paper systematically reviews the application of mainstream segmentation models-represented by encoder-decoder architectures, multi-scale feature fusion, and attention mechanisms-in this field. It provides an in-depth analysis of the technical characteristics, performance strengths and limitations, and applicable scenarios of various algorithms. By summarizing experimental findings and offering insights, this study aims to provide theoretical reference and forward-looking guidance for promoting the practical development and intelligent advancement of structural health monitoring technologies.

Downloads

Download data is not yet available.

References

[1] WANG Lu. Research on Semi-automatic Construction of Concrete Crack Semantic Segmentation Dataset and Model Performance[D]. Dalian Jiaotong University, 2025. DOI:10.26990/d.cnki.gsltc.2025.000826.

[2] FENG Jingyi, LIANG Hui, QI Zhiyong, et al. Concrete Crack Semantic Segmentation Model Integrating Multi-scale Features and Attention Mechanism[J]. Journal of Hydroelectric Engineering, 2025, 44(09): 114-124.

[3] PENG Y P, ZHANG R F, LIU Y H, et al. Lightweight concrete crack segmentation algorithm incorporating feature interaction and attention mechanisms[J]. Journal of Optoelectronics·Laser, 2025, 36(07): 722-732. DOI:10.16136/j.joel.2025.07.0070.

[4] YANG H L, CHEN J J, PAN Y. Research on automatic annotation and segmentation method for dam surface concrete cracks using Swin-Unet[J]. Journal of Hydroelectric Engineering, 2024, 43(12): 23-33.

[5] WANG Jianxin, LI Linyi, WANG Jin, et al. Pavement Crack Segmentation Method Integrating Frequency Domain Mamba and Recursively Gated Transformer[J/OL]. China Journal of Highway and Transport, 1-17[2025-11-15]. Link.

[6] DAI Shaosheng, MAO Xinghua, YU Zi'an. Image Processing-Based Method for Pavement Crack Feature Extraction[J]. Semiconductor Optoelectronics, 2024, 45(03): 508-514. DOI:10.16818/j.issn1001-5868.2024010304.

[7] TANG Y, DONG S J, LIU C, et al. A pavement crack segmentation algorithm based on improved SegFormer network[J]. Journal of Shaanxi University of Science & Technology, 2024, 42(03): 166-173. DOI:10.19481/j.cnki.issn2096-398x.2024.03.016.

[8] SONG Fengquan. Research on Semantic Segmentation Method for Concrete Bridge Cracks Based on Multi-scale Features[D]. Chongqing Jiaotong University, 2025. DOI:10.27671/d.cnki.gcjtc.2025.000729.

[9] YAO Yukai, GUO Baoyun, LI Cailin, et al. Research on Bridge Crack Segmentation Algorithm Based on Improved Deeplabv3+[J]. Journal of Shandong University of Technology(Natural Science Edition), 2024, 38(02): 21-26. DOI:10.13367/j.cnki.sdgc.2024.02.008.

[10] HU Wenkui, DENG Hui, FU Zhixu, et al. Bridge Crack Segmentation and Measurement Method Based on Fully Convolutional Neural Network[J]. Industrial Construction, 2022, 52(04): 192-201+218. DOI:10.13204/j.gyjzG21053111.