DWCF-YOLO: tunnel leakage detection and segmentation based on multi-level wavelet transform convolution

Abstract

Subway tunnels are essential components of urban transportation systems, and water leakage can threaten their structural safety. Accurate detection and segmentation of leakage regions are therefore critical. This study develops a tunnel water leakage dataset from point-cloud data collected using a mobile measurement platform and converts the point clouds into grayscale images. We propose DWCF-YOLO, an enhanced YOLOv11-based network for water leakage detection and segmentation. The model incorporates multi-level wavelet transform convolution to extract multi-frequency features and promote local–global feature fusion, improving the identification of leakage areas at different scales. In addition, content-guided attention provides channel- and spatial-directional guidance for multi-scale feature maps, enhancing sensitivity to leakage regions. Experimental results show that DWCF-YOLO achieves 96.8% accuracy, 97.8% recall, 98.4% F1-score, 98.9% precision, 96.8% mIoU, and 97.8% mPA. Compared with mainstream models, it obtains the best performance in four of the six evaluation metrics, demonstrating its effectiveness and reliability for tunnel water leakage detection.

Keywords

Subway tunnel / Water leakage detection / Image segmentation / Multi-source feature fusion

References

[1] Jiang J, Shen Y, Wang J, Wang J, Huang J, Fu S, Guo K, Ferreira V. Advances and challenges in water leakage detection techniques for shield tunnels: a comprehensive review. Measurement, 2026, 257: 118763

[2] Wu Y, Hu M, Xu G, Zhou X, Li Z (2019) Detecting leakage water of shield tunnel segments based on Mask R-CNN. In: 2019 IEEE International Conference on Architecture, Construction, Environment and Hydraulics (ICACEH) 25–28. https://doi.org/10.1109/ICACEH48424.2019.9042088

[3] Xue Y, Cai X, Shadabfar M, Shao H, Zhang S. Deep learning-based automatic recognition of water leakage area in shield tunnel lining. Tunnell Underground Space Technol, 2020, 104 103524

[4] Huang H, Cheng W, Zhou M, Chen J, Zhao S. Towards automated 3d inspection of water leakages in shield tunnel linings using mobile laser scanning data. Sensors (Switzerland), 2020, 20(221-23

[5] Xiong L, Zhang D, Zhang Y. Water leakage image recognition of shield tunnel via learning deep feature representation. J Vis Com Image Rep, 2020, 71: 102708

[6] Cheng X, Hu X, Tan K, Wang L, Yang L. Automatic Detection of Shield Tunnel Leakages Based on Terrestrial Mobile LiDAR Intensity Images Using Deep Learning. IEEE Access, 2021, 9: 55300-55310

[7] Zhao S, Zhang D, Chen J, Huang H, Shadabfar M (2021) Deep learning-based classification and instance segmentation of leakage-area and scaling images of shield tunnel linings. Struct Control Health Monitor 28(6). https://doi.org/10.1002/stc.2732

[8] Guo Z, Wei J, Sun H, Zhong R, Ji C. Enhanced Water Leakage Detection in Shield Tunnels Based on Laser Scanning Intensity Images Using RDES-Net. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2024, 17: 5680-5690

[9] Wang S, Mo J, Xu L, Zheng X (2024) Lightweight tunnel leakage water detection algorithm based on YOLOv8n. p 259–263. https://doi.org/10.1109/CCSSTA62096.2024.10691725

[10] Chen J, Xu X, Jeon G, Camacho D, He B-G. WLR-Net: An improved YOLO-V7 with edge constraints and attention mechanism for water leakage recognition in the tunnel. IEEE Transactions on Emerging Topics in Computational Intelligence, 2024, 8(4): 3105-3116

[11] Chen Q, Kang Z, Cao Z, Xie X, Guan B, Pan Y, Chang J. Combining Cylindrical Voxel and Mask R-CNN for Automatic Detection of Water Leakages in Shield Tunnel Point Clouds. REMOTE SENSING, 2024, 16(5): 896

[12] Wang P, Shi G. Image segmentation of tunnel water leakage defects in complex environments using an improved Unet model. Sci Rep, 2024, 14(1): 1-13

[13] Wang W, Xu X, Yang H. Intelligent detection of tunnel leakage based on improved mask R-CNN. Symmetry, 2024, 16(6): 709

[14] Chen J, Yu X, Liu S, Chen T, Wang W, Jeon G, He B. Tunnel SAM adapter: Adapting segment anything model for tunnel water leakage inspection. Geohazard Mechanics, 2024, 2(1): 29-36

[15] Jiang J, Shen Y, Wang J, Zang Y, Wu W, Wang J, Li J, Ferreira V. Boosted bagging: A hybrid ensemble deep learning framework for point cloud semantic segmentation of shield tunnel leakage. Tunnell Underground Space Technol, 2025, 164: 106842

[16] Liu Z, Gao X, Yang Y, Xu L, Wang S, Chen N, Wang Z, Kou Y. EDT-Net: A Lightweight Tunnel Water Leakage Detection Network Based on LiDAR Point Clouds Intensity Images. IEEE J Select Topics Appl Earth Observ Remote Sens, 2025, 18: 7334-7346

[17] Liang HY, Shen SL, Zhou A, Zhao WW. Automated detection and quantification of leakage areas in shield tunnel linings using laser scanning data and deep learning network. Eng Appl Artif Intell, 2025, 160 111930

[18] Yang R, Zhang H. Tunnel-YOLO: An improved You Only Look Once algorithm for real-time shield tunnel lining leakage detection. Eng Appl Artif Intell, 2025, 162 112403

[19] Zhang A, Huang J, Sun Z, Duan J, Zhang Y, Shen Y (2025) Leakage detection in subway tunnels using 3D point cloud data: integrating intensity and geometric features with XGBoost classifier. Sensors (14248220) 25(14):4475. https://doi.org/10.3390/s25144475

[20] Li W, Duan H, Zhang Q, Liang J, Duan W, Zhang K, et al.. Development of a Mobile Laser Measurement System for Subway Tunnel Deformation Detection. Sensors, 2025, 25(2): 356

[21] Yue Z, Sun H, Zhong R, Du L. Method for Tunnel Displacements Calculation Based on Mobile Tunnel Monitoring System. Sensors, 2021, 21(13): 4407

[22] Sun H, Xu Z, Yao L, Zhong R, Du L, Wu H (2020) Tunnel monitoring and measuring system using mobile laser scanning: design and deployment. Remote Sensing 12(4):730. https://doi.org/10.3390/rs12040730

[23] Sun H, Liu S, Zhong R, Du L. Cross-Section Deformation Analysis and Visualization of Shield Tunnel Based on Mobile Tunnel Monitoring System. Sensors, 2020, 20(4): 1006

[24] Ji C, Sun H, Zhong R, Li J, Han Y. Precise Positioning Method of Moving Laser Point Cloud in Shield Tunnel Based on Bolt Hole Extraction. Remote Sensing, 2022, 14(19): 4791

[25] Jocher G, Qiu J (2024) Ultralytics YOLO11 (Version 11.0.0) [Computer software]. GitHub. https://github.com/ultralytics/ultralytics. License: AGPL-3.0

[26] Finder SE, Amoyal R, Treister E, Freifeld O (2024) Wavelet convolutions for large receptive fields. In: Lecture Notes in Computer Science, pp 363–380. Springer Nature Switzerland. https://doi.org/10.1007/978-3-031-72949-2_21

[27] Chen Z, He Z, Lu ZM. DEA-Net: Single Image Dehazing Based on Detail-Enhanced Convolution and Content-Guided Attention. IEEE Trans Image Process, 2024, 33: 1002-1015

Funding

Key R&D Program of Xuzhou City(KC23295)

Graduate Innovation Program of China University of Mining and Technology(KYCX25_3050)

National Natural Science Foundation of China(U22A20569)

Rights & permissions

The Author(s)