Author(s)

Fang Wang

Affiliation(s)

School of Computer and Artificial Intelligence, Beijing Technology and Business University, Beijing 100048, China.

Corresponding Author

Fang Wang

ABSTRACT

This paper addresses the problem of low segmentation accuracy of cracks and joints in complex scenarios and proposes an improved model, ViR-Deeplabv3+, based on DeepLabv3+. First, the model replaces the traditional backbone network with the Vision Transformer (ViT) with global perception ability. This enables the model to no longer be limited to extracting local information when processing image data, but to capture the global context features of the image more efficiently, thereby enhances subsequent segmentation tasks. Secondly, residual connections between the ViT and the Hollow Space Pyramid Pooling (ASPP) module are ingeniously introduced. The design concept of residual connection effectively solves problems such as the gradient vanishing problem, ensuring that the rich feature information from ViT can be smoothly and unobstructedly transmitted to the ASPP module for further fusion and mining of multi-scale features. Finally, we conducted model training and ablation experiments based on the self-built dataset (including crack and seam samples). The results showed that the mean intersection and union ratio (mIoU) of ViR-Deeplabv3+ reached 75.27%, which was 2.97% higher than that of the baseline model Deeplabv3+. This scheme provides an effective solution for precisely detecting and segmenting cracks and joints in complex scenarios, and has important practical application value.

KEYWORDS

Image segmentation; Crack; Seam; ViT; Residual connection

CITE THIS PAPER

Fang Wang. Joint segmentation model for cracks and joints based on DeepLabv3+. World Journal of Information Technology. 2025, 3(3): 33-40. DOI: https://doi.org/10.61784/wjit3040.

REFERENCES

[1] Ma Jing, Liu Lin, Lv Suyan. Monitoring and evaluation technology of cracks in concrete bridges. Comprehensive Corrosion Control, 2025, 39(4): 185-188. 

[2] Pratibha K, Mishra M, Ramana G V, et al. Deep learning-based yolo network model for detecting surface cracks during structural health monitoring//International Conference on Structural Analysis of Historical Constructions. Cham: Springer Nature Switzerland, 2023, 179-187.

[3] Marin B, Brown K, Erden M S, et al. Automated masonry crack detection with faster R-CNN//2021 IEEE 17th International Conference on Automation Science and Engineering (CASE), Lyon, France, 2021, 333-340. DOI: 10.1109/CASE49439.2021.9551683.

[4] Wang Yanhua, He Junze, Zhang Mingzhou, et al. Complex-environment concrete crack recognition based on SSD and pruned neural network. Journal of Southeast University (English Edition), 2023, 39(4): 393-399.

[5] Lau Stephen L H, Chong Edwin K P, Yang Xu, et al. Automated pavement crack segmentation using U-Net-based convolutional neural network. IEEE Access, 2020, 8, 114892-114899.

[6] Attard L, Debono C L, Valentino G, et al. Automatic crack detection using mask R-CNN[C]//2019 11th international symposium on image and signal processing and analysis (ISPA), Dubrovnik, Croatia, 2019, 152-157. DOI: 10.1109/ISPA.2019.8868619.

[7] Yao Yukai, Guo Baoyun, Li Cailin, et al. Bridge crack segmentation algorithm based on improved Deeplabv3+. Journal of Shandong University of Technology (Natural Science Edition), 2024, 38(2): 21-26.

[8] Chen Liang-Chieh, Zhu Yukun, Papandreou G, et al. Encoder-decoder with atrous separable convolution for semantic image segmentation. 2018. DOI: https://doi.org/10.48550/arXiv.1802.02611.

[9] Torres-Acosta A A, Martínez-Madrid M. Residual life of corroding reinforced concrete structures in marine environment. Journal of Materials in Civil Engineering, 2003, 15(4): 344-353.

[10] WANG Y, ZHANG H. Impact of joint-crack mixed datasets on semantic segmentation models. IEEE Transactions on Image Processing, 2022, 31(5): 2345-2356.