EDA-TCNet: A dual-attention enhanced network for precise point cloud segmentation in tunnel construction

Abstract

To enhance the accuracy of point cloud semantic segmentation in tunnel face construction areas, this study proposes a novel model named enhanced dual attention-tunnel construction net (EDA-TCNet). EDA-TCNet introduces a 3D enhanced dual attention module (EDAM), which employs a parallel channel and spatial attention mechanism to strengthen the model’s focus on critical features. Additionally, a loss function named CELDAM is designed, combining cross-entropy loss and label-distribution-aware margin loss to effectively address data imbalance issues and improve the prediction capability for minority classes. Experiments conducted on three ongoing tunnel projects in Northwest China demonstrate that EDA-TCNet achieves a mean intersection over union (mIoU) of 0.8816 and an overall accuracy (OA) of 0.9406 on the test set. Compared to PointNet, PointNet++, DGCNN, and PointMLP, EDA-TCNet improves mIoU by 18.20%, 3.00%, 8.61%, and 32.23%, and OA by 15.98%, 1.74%, 5.48%, and 22.38%, respectively. Furthermore, the optimization of the balancing coefficient μ in CELDAM further enhances the model’s generalization capability. In conclusion, EDA-TCNet demonstrates exceptional performance in point cloud semantic segmentation tasks for tunnel construction areas and shows great potential for engineering applications.

Keywords

Point cloud / Semantic segmentation / Deep learning / Enhanced dual attention / Tunnel construction

Data availability

The data that used in this study are available on https://github.com/ROCKMASSPX/EDATCNet.

CRediT authorship contribution statement

Xin Peng: Writing - review & editing, Writing - original draft, Visualization, Validation, Methodology, Investigation, Conceptualization. Mingnian Wang: Writing - review & editing, Supervision, Resources, Project administration. Bingxu Huang: Validation, Data curation. Honglin Shen: Validation. Hao Zhong: Validation.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgement

This work was supported by the National Natural Science Foundation of China (Grant Nos. 52378411 and 51878567).

References

[1] Attard, L., Debono, C. J., Valentino, G., & Di Castro, M. (2018). Tunnel inspection using photogrammetric techniques and image processing: A review. Isprs Journal of Photogrammetry and Remote Sensing, 144, 180-188.

[2] Barton, N. (1988). Rock mass classification and tunnel reinforcement selection using the q-system.

[3] Battulwar, R., Zare-Naghadehi, M., Emami, E., & Sattarvand, J. (2021). A state-of-the-art review of automated extraction of rock mass discontinuity characteristics using three-dimensional surface models. Journal of Rock Mechanics and Geotechnical Engineering, 13(4), 920-936.

[4] Chen, J., Yang, T., Zhang, D., Huang, H., & Tian, Y. (2021a). Deep learning based classification of rock structure of tunnel face. Geoscience Frontiers, 12(1), 395-404.

[5] Chen, J., Zhang, D., Huang, H., Shadabfar, M., Zhou, M., & Yang, T. (2020). Image-based segmentation and quantification of weak interlayers in rock tunnel face via deep learning. Automation in Construction, 120, 103371.

[6] Chen, J., Zhou, M., Zhang, D., Huang, H., & Zhang, F. (2021b). Quantification of water inflow in rock tunnel faces via convolutional neural network approach. Automation in Construction, 123, 103526.

[7] Cui, H., Li, J., Mao, Q., Hu, Q., Dong, C., & Tao, Y. (2024). STSD: A large-scale benchmark for semantic segmentation of subway tunnel point cloud. Tunnelling and Underground Space Technology, 150, 105829.

[8] Daraei, A., & Zare, S. (2018). Prediction of overbreak depth in Ghalaje road tunnel using strength factor. International Journal of Mining Science and Technology, 28(4), 679-684.

[9] Drews, T., Miernik, G., Anders, K., Höfle, B., Profe, J., Emmerich, A., & Bechstädt, T. (2018). Validation of fracture data recognition in rock masses by automated plane detection in 3D point clouds. International Journal of Rock Mechanics and Mining Sciences, 109, 19-31.

[10] Fang, S., Xu, D., Zhao, Z., Song, Q., & Gui, W. (2024). Shotcrete flatness evaluation of initial linings based on vehicular LiDAR scanning. Automation in Construction, 164, 105475.

[11] Fodera`, G., Voza, A., Barovero, G., Tinti, F., & Boldini, D. (2020). Factors influencing overbreak volumes in drill-and-blast tunnel excavation. A statistical analysis applied to the case study of the Brenner Base Tunnel-BBT. Tunnelling and Underground Space Technology, 105, 103475.

[12] García-Luna, R., Senent, S., Jurado-Piña, R., & Jimenez, R. (2019). Structure from Motion photogrammetry to characterize underground rock masses: Experiences from two real tunnels. Tunnelling and Underground Space Technology, 83, 262-273.

[13] Gischig, V., Amann, F., Moore, J., Loew, S., Eisenbeiss, H., & Stempfhuber, W. (2011). Composite rock slope kinematics at the current Randa instability, Switzerland, based on remote sensing and numerical modeling. Engineering Geology, 118(1-2), 37-53.

[14] Gong, J. F., Wang, W., Wang, F., Yang, C. X., & Yuan, Y. (2024). Statistics of China’s railway tunnels by the end of 2023 and overview of tunnels of key new projects in 2023. Tunnel Construction, 44(2), 377-392 (in Chinese).

[15] Grandio, J., Riveiro, B., Lamas, D., & Arias, P. (2023). Multimodal deep learning for point cloud panoptic segmentation of railway environments. Automation in Construction, 150, 104854.

[16] Grandio, J., Riveiro, B., Soilán, M., & Arias, P. (2022). Point cloud semantic segmentation of complex railway environments using deep learning. Automation in Construction, 141, 104425.

[17] Han, J.-Y., Guo, J., & Jiang, Y.-S. (2013). Monitoring tunnel deformations by means of multi-epoch dispersed 3D LiDAR point clouds: An improved approach. Tunnelling and Underground Space Technology, 38, 385-389.

[18] He, C., & Wang, B. (2013). Research progress and development trends of highway tunnels in China. Journal of Modern Transportation, 21(4), 209-223.

[19] Hong, Z., Tao, M., Cui, X., Wu, C., & Zhao, M. (2023). Experimental and numerical studies of the blast-induced overbreak and underbreak in underground roadways. Underground Space, 8, 61-79.

[20] Ji, A., Zhang, L., Fan, H., Xue, X., & Dou, Y. (2023). Dual attention-based deep learning network for multi-class object semantic segmentation of tunnel point clouds. Automation in Construction, 156, 105131.

[21] Jiang, M., Wu, Y., Zhao, T., Zhao, Z., & Lu, C. (2018). Pointsift: A sift-like network module for 3d point cloud semantic segmentation. arXiv preprint arXiv:1807.00652.

[22] Kang, J., Chen, N., Li, M., Mao, S., Zhang, H., Fan, Y., & Liu, H. (2023). A point cloud segmentation method for dim and cluttered underground tunnel scenes based on the segment anything model. Remote Sensing, 16(1), 97.

[23] Kemeny, J., & Post, R. (2003). Estimating three-dimensional rock discontinuity orientation from digital images of fracture traces. Computers & Geosciences, 29(1), 65-77.

[24] Lee, J. S., Park, J., & Ryu, Y.-M. (2021). Semantic segmentation of bridge components based on hierarchical point cloud model. Automation in Construction, 130, 103847.

[25] Li, X., Sun, X., Meng, Y., Liang, J., Wu, F., & Li, J. (2019). Dice loss for data-imbalanced NLP tasks. arXiv preprint arXiv:1911.02855.

[26] Li, Y., Bu, R., Sun, M., Wu, W., Di, X., & Chen, B. (2018). Pointcnn: Convolution on x-transformed points. In Advances in neural information processing systems (pp.31).

[27] Lin, T. (2017). Focal Loss for Dense Object Detection. arXiv preprint arXiv:1708.02002.

[28] Lin, W., Sheil, B., Zhang, P., Zhou, B., Wang, C., & Xie, X. (2024). Seg2Tunnel: A hierarchical point cloud dataset and benchmarks for segmentation of segmental tunnel linings. Tunnelling and Underground Space Technology, 147, 105735.

[29] Ma, X., Qin, C., You, H., Ran, H., & Fu, Y. (2022). Rethinking network design and local geometry in point cloud: A simple residual MLP framework. arXiv preprint arXiv:2202.07123.

[30] Mahtab, M., Rossler, K., Kalamaras, G., & Grasso, P. (1997). Assessment of geological overbreak for tunnel design and contractual claims. International Journal of Rock Mechanics and Mining Sciences, 34(3-4), 185, e181-185. e113.

[31] Monsalve, J. J., Baggett, J., Bishop, R., & Ripepi, N. (2019). Application of laser scanning for rock mass characterization and discrete fracture network generation in an underground limestone mine. International Journal of Mining Science and Technology, 29(1), 131-137.

[32] Nuttens, T., Stal, C., De Backer, H., Schotte, K., Van Bogaert, P., & De Wulf, A. (2014). Methodology for the ovalization monitoring of newly built circular train tunnels based on laser scanning: Liefkenshoek Rail Link (Belgium). Automation in Construction, 43, 1-9.

[33] Peng, X., Wang, M., Huang, B., & Lin, P. (2024). Efficient automated method for characterizing discontinuities in tunnel face rock mass point clouds. Tunnelling and Underground Space Technology, 154, 106117.

[34] Phan, A. V., Le Nguyen, M., Nguyen, Y. L. H., & Bui, L. T. (2018). Dgcnn: A convolutional neural network over large-scale labeled graphs. Neural Networks, 108, 533-543.

[35] Priest, S. D. (1993). Discontinuity analysis for rock engineering. Dordrecht: Springer, Netherlands.

[36] Qi, C. R., Su, H., Mo, K., & Guibas, L. J. (2017a). Pointnet: Deep learning on point sets for 3d classification and segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp.652-660).

[37] Qi, C. R., Yi, L., Su, H., & Guibas, L. J. (2017b). Pointnet++: Deep hierarchical feature learning on point sets in a metric space. In Advances in Neural Information Processing Systems (pp.30).

[38] Roca-Pardiñas, J., Argüelles-Fraga, R., de Asís López, F., & Ordóñez, C. (2014). Analysis of the influence of range and angle of incidence of terrestrial laser scanning measurements on tunnel inspection. Tunnelling and Underground Space Technology, 43, 133-139.

[39] Singh, S. K., Banerjee, B. P., Lato, M. J., Sammut, C., & Raval, S. (2022). Automated rock mass discontinuity set characterisation using amplitude and phase decomposition of point cloud data. International Journal of Rock Mechanics and Mining Sciences, 152, 105072.

[40] Singh, S. K., Banerjee, B. P., & Raval, S. (2023). A review of laser scanning for geological and geotechnical applications in underground mining. International Journal of Mining Science and Technology, 33(2), 133-154.

[41] Singh, S. K., Raval, S., & Banerjee, B. P. (2021). Automated structural discontinuity mapping in a rock face occluded by vegetation using mobile laser scanning. Engineering Geology, 285, 106040.

[42] Singh, S. P., & Xavier, P. (2005). Causes, impact and control of overbreak in underground excavations. Tunnelling and Underground Space Technology, 20(1), 63-71.

[43] Soilán, M., Nóvoa, A., Sánchez-Rodríguez, A., Riveiro, B., & Arias, P. (2020). Semantic segmentation of point clouds with pointnet and kpconv architectures applied to railway tunnels. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 2, 281-288.

[44] Sun, X., He, L., Jiang, H., Li, R., Mao, W., Zhang, D., Majeed, Y., Andriyanov, N., Soloviev, V., & Fu, L. (2024). Morphological estimation of primary branch length of individual apple trees during the deciduous period in modern orchard based on PointNet++. Computers and Electronics in Agriculture, 220, 108873.

[45] Wang, W., Yu, R., Huang, Q., & Neumann, U. (2018). Sgpn: Similarity group proposal network for 3d point cloud instance segmentation. Proceedings of the IEEE conference on computer vision and pattern recognition.

[46] Woo, S., Park, J., Lee, J.-Y., & Kweon, I. S. (2018). Cbam: Convolutional block attention module. Proceedings of the European conference on computer vision (ECCV).

[47] Xie, Y., Tian, J., & Zhu, X. X. (2020). Linking points with labels in 3D: A review of point cloud semantic segmentation. IEEE Geoscience and remote sensing magazine, 8(4), 38-59.

[48] Yang, X., Lu, B., Jing, Q., Li, Y., & Zhou, Y. (2023). Macrotexture deterioration for micromilled tunnel concrete pavement using 3D laser data. Measurement Science and Technology, 34(6), 065001.

[49] Yin, C., Wang, B., Gan, V. J., Wang, M., & Cheng, J. C. (2021). Automated semantic segmentation of industrial point clouds using ResPointNet++. Automation in Construction, 130, 103874.

[50] Zeng, Y., & Wang, Z. (2014). A policy analysis on challenges and opportunities of population/household aging in China. Journal of Population Ageing, 7(4), 255-281.

[51] Zhang, L., & Wang, H. (2021). A novel segmentation method for cervical vertebrae based on PointNet++ and converge segmentation. Computer Methods and Programs in Biomedicine, 200, 105798.

[52] Zhang, L., Wei, Z., Xiao, Z., Ji, A., & Wu, B. (2024). Dual hierarchical attention-enhanced transfer learning for semantic segmentation of point clouds in building scene understanding. Automation in Construction, 168, 105799.

[53] Zhang, Z., Ji, A., Wang, K., & Zhang, L. (2022). UnrollingNet: An attention-based deep learning approach for the segmentation of large-scale point clouds of tunnels. Automation in Construction, 142, 104456.

[54] Zhang, Z., Ji, A., Zhang, L., Xu, Y., & Zhou, Q. (2023). Deep learning for large-scale point cloud segmentation in tunnels considering causal inference. Automation in Construction, 152, 104915.

[55] Zhao, H., Jiang, L., Jia, J., Torr, P. H., & Koltun, V. (2021). Point transformer. Proceedings of the IEEE/CVF international conference on computer vision.

[56] Zhou, J.-W., Chen, J.-L., & Li, H.-B. (2024). An optimized fuzzy K-means clustering method for automated rock discontinuities extraction from point clouds. International Journal of Rock Mechanics and Mining Sciences, 173, 105627.

[57] Zhou, Y., Ji, A., Zhang, L., & Xue, X. (2023). Attention-enhanced sampling point cloud network (ASPCNet) for efficient 3D tunnel semantic segmentation. Automation in Construction, 146, 104667.

[58] Zhu, H., Yan, J., & Liang, W. (2019). Challenges and development prospects of ultra-long and ultra-deep mountain tunnels. Engineering, 5(3), 384-392.