Machine Learning Theory and Practice, 2025, 5(1); doi: 10.38007/ML.2025.050103.
Qiang Zhou, Jie Gou
Sichuan Shudao New Energy Technology Development Co., Ltd, , Chengdu, Sichuan, China 610041
Southwest Jiaotong University, Sichuan, China 610031
Because the reliability of manual crack detection method is low, the cost is high, and it is greatly affected by subjectivity. This paper studies the image recognition of pavement cracks by using digital image technology. This method has the advantages of high accuracy, high flexibility, high speed and low cost, which is not affected by human experience The image processing of concrete cracks using digital image processing technology includes image grayscale semantic segmentation convolution neural network structural crack recognition and so on In this paper, we use Pytorch platform, Python programming language and U-Net neural network structure to build a model training platform to carry out model training independently, and summarize the processing technology of pavement crack image to provide reference for researchers in this field This method can quickly and effectively identify the type, size, shape and other information of cracks in the image.
Pavement crack identification; Pytorch; crack; Deep learning; Unet; Python; Semantic segmentation
Qiang Zhou, Jie Gou. Crack Identification Based on Pytorch and U-Net Neural Network. Machine Learning Theory and Practice (2025), Vol. 5, Issue 1: 27-33. https://doi.org/10.38007/ML.2025.050103.
[1] Niu, K. (2023). Analysis of the causes and prevention measures of highway pavement cracks. Transportation World, Z1, 77-79+82.
[2] Li, H., Wu, Z., Nie, J., et al. (2020). Airport pavement crack automatic detection algorithm based on depth images. Journal of Traffic and Transportation Engineering, 20(06), 250-260. DOI:10.19818/j.cnki.1671-1637.2020.06.022.
[3] Chen, T., Cai, Z., Zhao, X., et al. (2020). Pavement crack detection and recognition using the architecture of SegNet. Journal of Industrial Information Integration, 18, 100144. DOI:10.1016/j.jii.2020.100144.
[4] Guo, C., Gao, W., Zhou, D. (2024). Research on road surface crack detection based on SegNet network. Journal of Engineering and Applied Science, 71(1), 54. DOI:10.1186/s44147-024-00391-0.
[5] Tokhmechi, B., Memarian, H., Noubari, H. A., et al. (2009). A novel approach proposed for fractured zone detection using petrophysical logs. Journal of Geophysics and Engineering, 6(4), 365-373.
[6] Gao, L., Rao, F., Yang, Z., et al. (2024). Research on intelligent recognition technology for road surface cracks based on semantic segmentation. China & Foreign Highway, 44(5), 241-247. DOI:10.14048/j.issn.1671-2579.2024.05.027.