Composite Loss-Enhanced DeepLabV3+ for Crack Segmentation in Photogrammetric Mining Imagery

Yesri Elva; Iskandar  Fitri; Syafri Arlis

doi:10.29207/resti.v10i4.7807

Authors

Yesri Elva Universitas Putra Indonesia YPTK Padang
Iskandar Fitri Universitas Putra Indonesia YPTK Padang
Syafri Arlis Universitas Putra Indonesia YPTK Padang

DOI:

https://doi.org/10.29207/resti.v10i4.7807

Keywords:

DeepLabV3, Focal Tversky Loss, land cracks, ResNet-50, segmentation

Abstract

Crack detection in photogrammetric mining imagery remains challenging due to severe class imbalance, thin and irregular crack morphology, and complex soil textures that hinder accurate boundary delineation. This study proposes a DeepLabV3+ semantic segmentation model with a ResNet-50 backbone trained using a Composite Loss that integrates Focal Tversky Loss, Weighted Cross-Entropy Loss, and Boundary Loss into a unified multi-objective framework. The dataset comprised 425 photogrammetric crack images acquired from open-pit mining sites, divided into 380 training and 45 test images. Data augmentation, including horizontal flip, vertical flip, channel shuffle, and random brightness and contrast adjustment, was applied to improve model generalization under varying image conditions. The proposed configuration was evaluated against Default Loss, Binary Cross-Entropy Loss, and Weighted Cross-Entropy Loss. The model achieved a Global Accuracy of 0.9808, Mean Intersection over Union of 0.7890, and Mean Boundary F1-Score of 0.8642, outperforming all baseline configurations. Boundary fidelity showed the largest improvement, with a Mean BFScore gain of 27.3% over Default Loss, demonstrating the effectiveness of Boundary Loss for precise crack contour delineation. The trained model was also deployed in a GUI-based system supporting automated crack analysis and PDF report generation for practical field application.

Downloads

Download data is not yet available.

References

[1] C. Cacciuttolo, V. Guzmán, P. Catriñir, E. Atencio, S. Komarizadehasl, and J. A. Lozano-Galant, “Low-Cost Sensors Technologies for Monitoring Sustainability and Safety Issues in Mining Activities: Advances, Gaps, and Future Directions in the Digitalization for Smart Mining,” 2023. doi: 10.3390/s23156846.

[2] Y. Tang, B. Yang, X. Zhao, and C. Yang, “Structural and Fractal Analysis of Soil Cracks Due to the Roots of Setaria Viridis,” Fractal and Fractional, vol. 7, no. 1, 2023, doi: 10.3390/fractalfract7010019.

[3] Y. Wu et al., “Failure mechanism and early warning of an excavation-induced soil landslide,” Front. Earth Sci. (Lausanne)., vol. 12, 2024, doi: 10.3389/feart.2024.1410011.

[4] A. Mardanshahi, A. Sreekumar, X. Yang, S. K. Barman, and D. Chronopoulos, “Sensing Techniques for Structural Health Monitoring: A State-of-the-Art Review on Performance Criteria and New-Generation Technologies,” 2025. doi: 10.3390/s25051424.

[5] M. Konieczna-Fuławka et al., “Autonomous Mobile Inspection Robots in Deep Underground Mining—The Current State of the Art and Future Perspectives,” 2025. doi: 10.3390/s25123598.

[6] N. F. Alkayem et al., “Co-CrackSegment: A New Collaborative Deep Learning Framework for Pixel-Level Semantic Segmentation of Concrete Cracks,” Mathematics, vol. 12, no. 19, 2024, doi: 10.3390/math12193105.

[7] T. Lee, J. H. Kim, S. J. Lee, S. K. Ryu, and B. C. Joo, “Improvement of Concrete Crack Segmentation Performance Using Stacking Ensemble Learning,” Applied Sciences (Switzerland), vol. 13, no. 4, 2023, doi: 10.3390/app13042367.

[8] P. Shokri, M. Shahbazi, and J. Nielsen, “Semantic Segmentation and 3D Reconstruction of Concrete Cracks,” Remote Sens. (Basel)., vol. 14, no. 22, 2022, doi: 10.3390/rs14225793.

[9] Y. Li, “The research on landslide detection in remote sensing images based on improved DeepLabv3+ method,” Sci. Rep., vol. 15, no. 1, 2025, doi: 10.1038/s41598-025-92822-y.

[10] H. Liu, Y. Chen, R. Wang, M. Li, and Z. Li, “MFA-Deeplabv3+: an improved lightweight semantic segmentation algorithm based on Deeplabv3+,” Complex and Intelligent Systems, vol. 11, no. 10, 2025, doi: 10.1007/s40747-025-02028-y.

[11] X. Jia, Y. Wang, and Z. Wang, “Fatigue Crack Detection Based on Semantic Segmentation Using DeepLabV3+ for Steel Girder Bridges,” Applied Sciences (Switzerland), vol. 14, no. 18, 2024, doi: 10.3390/app14188132.

[12] J. Yan, X. Li, and X. Zhou, “Intelligent Identification of Liquid Aluminum Leakage in Deep Well Casting Production Based on Image Segmentation,” Applied Sciences (Switzerland), vol. 14, no. 13, 2024, doi: 10.3390/app14135470.

[13] A. P, V. P, S. Kumar S, and J. Naidu K, “Adaptive multilevel attention deeplabv3+ with heuristic based frame work for semantic segmentation of aerial images using improved golden jackal optimization algorithm,” Results in Engineering, vol. 24, 2024, doi: 10.1016/j.rineng.2024.103164.

[14] Y. Wang, L. Yang, X. Liu, and P. Yan, “An improved semantic segmentation algorithm for high-resolution remote sensing images based on DeepLabv3+,” Sci. Rep., vol. 14, no. 1, 2024, doi: 10.1038/s41598-024-60375-1.

[15] Y. Fan, Z. Hu, Q. Li, Y. Sun, J. Chen, and Q. Zhou, “CrackNet: A Hybrid Model for Crack Segmentation with Dynamic Loss Function,” Sensors, vol. 24, no. 22, 2024, doi: 10.3390/s24227134.

[16] K. A. Gobena, M. Y. K. Rakib, F. B. Tesema, G. F. Asafa, and S. Ren, “LiteCrackSeg: A lightweight hybrid CNN–transformer for efficient crack segmentation,” PLoS One, vol. 21, no. 4, p. e0347765, Apr. 2026, doi: 10.1371/journal.pone.0347765.

[17] A. A. Sundaresan and A. A. Solomon, “Post-disaster flooded region segmentation using DeepLabv3+ and unmanned aerial system imagery,” Natural Hazards Research, vol. 5, no. 2, 2025, doi: 10.1016/j.nhres.2024.12.003.

[18] F. E. Oğuz and A. Alkan, “AI-enhanced interface for colonic polyp segmentation using DeepLabv3+ with comparative backbone analysis,” Biomed. Phys. Eng. Express, vol. 11, no. 1, 2025, doi: 10.1088/2057-1976/ada15f.

[19] A. F. Rogachev and I. S. Belousov, “Simulation of the of the DeepLabv3 neural network learning process for the agricultural fields segmentation,” Herald of Dagestan State Technical University. Technical Sciences, vol. 50, no. 3, 2023, doi: 10.21822/2073-6185-2023-50-3-142-149.

[20] R. Singh and N. Sharma, “Improving Brain Tumor Detection and Segmentation with DeepLabV3 and ResNet-50,” in 2024 5th IEEE Global Conference for Advancement in Technology, GCAT 2024, 2024. doi: 10.1109/GCAT62922.2024.10923939.

[21] H. Matich, J. Attmani, and H. Mousannif, “Enhancing crack detection and severity assessment in historical Tabiya basins using U-Net and adaptive thresholding,” Frontiers in Artificial Intelligence, vol. 9, 2026, doi: 10.3389/frai.2026.1741082.

[22] R. B. de Andrade, G. L. A. Mota, and G. A. O. P. da Costa, “Deforestation Detection in the Amazon Using DeepLabv3+ Semantic Segmentation Model Variants,” Remote Sens. (Basel)., vol. 14, no. 19, 2022, doi: 10.3390/rs14194694.

[23] X. Jia, Y. Wang, and Z. Wang, “Fatigue Crack Detection Based on Semantic Segmentation Using DeepLabV3+ for Steel Girder Bridges,” Applied Sciences, vol. 14, no. 18, Art. no. 8132, 2024, doi: 10.3390/app14188132.

[24] A. D. Salsabila, F. Fatimah, D. Darmini, and S. B. Kurniawan, “Deep Learning-Based Hippocampal Segmentation and MTA Classification Using U-Net with ResNet-50 Backbone,” Indonesian Journal of Electronics, Electromedical Engineering, and Medical Informatics, vol. 7, no. 4, 2025, doi: 10.35882/ijeeemi.v7i4.263.

[25] M. Phạm, “Landslide crack dataset,” Mendeley Data, ver. 1, 2022, doi: 10.17632/vd5fzm7d4z.1.

[26] E. A. Shamsabadi, S. M. H. Erfani, C. Xu, and D. Dias-da-Costa, “Efficient semi-supervised surface crack segmentation with small datasets based on consistency regularisation and pseudo-labelling,” Automation in Construction, vol. 158, Art. no. 105181, 2024, doi: 10.1016/j.autcon.2023.105181.

[27] J. Ju, H. Noh, Y. Wang, and D. G. Choi, “Class-Wise Adaptive Strategy for Semi Supervised Semantic Segmentation,” IEEE Access, vol. 12, 2024, doi: 10.1109/ACCESS.2024.3363628.

[28] W. Huang, Y. Shi, Z. Xiong, and X. X. Zhu, “Decouple and weight semi-supervised semantic segmentation of remote sensing images,” ISPRS Journal of Photogrammetry and Remote Sensing, vol. 212, 2024, doi: 10.1016/j.isprsjprs.2024.04.010.

[29] Z. Zhang et al., “Algorithm for pixel-level concrete pavement crack segmentation based on an improved U-Net model,” Scientific Reports, vol. 15, Art. no. 6553, 2025, doi: 10.1038/s41598-025-91352-x.

[30] F. Ketenci Çay, Ç. Yeşil, O. Çay, B. G. Yılmaz, F. H. Özçini, and D. İlgüy, “DeepLabv3 + method for detecting and segmenting apical lesions on panoramic radiography,” Clin. Oral Investig., vol. 29, no. 2, 2025, doi: 10.1007/s00784-025-06156-0.

[31] X. Chen, S. Wang, V. Dinavahi, L. Yang, D. Wu, and M. Shen, “Landslide Recognition Based on DeepLabv3+ Framework Fusing ResNet101 and ECA Attention Mechanism,” Applied Sciences (Switzerland), vol. 15, no. 5, 2025, doi: 10.3390/app15052613.

[32] Y. Liu, “DeepLabV3+ Based Mask R-CNN for Crack Detection and Segmentation in Concrete Structures,” International Journal of Advanced Computer Science and Applications, vol. 16, no. 1, 2025, doi: 10.14569/IJACSA.2025.0160142.

[33] F. Li, Y. Mou, Z. Zhang, Q. Liu, and S. Jeschke, “A novel model for the pavement distress segmentation based on multi-level attention DeepLabV3+,” Eng. Appl. Artif. Intell., vol. 137, 2024, doi: 10.1016/j.engappai.2024.109175.

[34] H. Ajra, M. A. Majid, M. S. Islam, and D. Abdullah, “Leaf Disease Detection in Plant Care using CNN Architecture: AlexNet and ResNet-50 Models,” Int. J. Adv. Sci. Eng. Inf. Technol., vol. 15, no. 1, 2025, doi: 10.18517/ijaseit.15.1.19944.

[35] V. Kumar and A. Rajput, “CIFAR-10 Object Detection Using ResNet-50 Model,” in Lecture Notes in Networks and Systems, 2025. doi: 10.1007/978-981-97-8836-1_26.