Automated Pixel-Level Concrete Defect Detection using U-Net Architecture: A Comparative Study with Clustering-Based Segmentation

Halifia Hendri, Larissa Navia Rani, Sofika Enggari, Agung Ramadhanu, Febri Hadi
Pages 1379-1392
Abstract Views 317
PDF Downloads 356
Crossref Citations -
DOI https://doi.org/10.47738/jads.v7i2.1298

Abstract


Concrete surface defect detection is a critical aspect of maintaining the integrity and safety of infrastructure in civil engineering. Traditional manual inspection methods are time-consuming, prone to human subjectivity, and often limited by physical accessibility, necessitating the development of robust automated solutions. This paper presents an automated pixel-level concrete surface defect detection system utilizing the U-Net deep learning architecture. The primary contribution and novelty of our approach lie in optimizing the network's encoder-decoder structure with skip connections to effectively capture both broad contextual features and precise spatial localization. This overcomes the critical limitations of existing traditional methods, which frequently struggle with complex concrete background textures, inherent noise, and uneven illumination. To validate our approach, the proposed U-Net model is systematically compared against a widely used baseline method, K-Means clustering combined with Gray-Level Co-occurrence Matrix (GLCM) texture analysis. The evaluation was conducted using a comprehensive dataset consisting of 1000 high-resolution concrete images. Experimental results reveal that the deep learning architecture vastly outperforms the traditional baseline. Specifically, the U-Net model achieved an outstanding F1-Score of 92.47%, a precision of 93.18%, and a mean Intersection over Union (mIoU) of 86.55%. In stark contrast, the K-Means and GLCM approach only yielded an F1-Score of 69.83% and an mIoU of 54.21%. These quantitative findings demonstrate that the proposed U-Net-based system not only successfully minimizes false segmentations but also provides a highly reliable, efficient, and scalable computational framework. Ultimately, this research delivers a practical solution that can be seamlessly integrated into continuous automated structural health monitoring systems, paving the way for safer and more proactive civil infrastructure management.

Keywords


Concrete Defect Detection; U-Net Architecture; Deep Learning; Semantic Segmentation; Structural Health Monitoring (SHM); Non-Destructive Testing (NDT); Construction Automation

Full Text:

PDF

References


C. Ferraris, G. Amprimo, and G. Pettiti, "Computer vision and image processing in structural health monitoring: Overview of recent applications," Signals, vol. 4, no. 3, pp. 539–574, Jul. 2023, doi: 10.3390/signals4030029.

D. T. H. O’Connor, M. A. N. Hassan, and B. A. Harvey, "Image-based structural health monitoring: A systematic review," Appl. Sci., vol. 13, no. 2, p. 968, Jan. 2023, doi: 10.3390/app13020968.

B. F. Spencer Jr., S.-H. Sim, R. E. Kim, and H. Yoon, "Advances in artificial intelligence for structural health monitoring: A comprehensive review," KSCE J. Civ. Eng., vol. 29, no. 3, p. 100203, 2025, doi: 10.1007/s12205-024-1002-3.

Y. Yu, J. Li, J. Li, and X. Xia, "Concrete crack detection based on attention mechanism and multi-scale feature fusion," Journal of Building Engineering, vol. 68, p. 106068, Jun. 2023, doi: 10.1016/j.jobe.2023.106068.

S. Guo, X. Xu, and Y. Zhu, "Illumination-invariant concrete crack detection using deep learning with synthetic data augmentation," Automation in Construction, vol. 158, p. 105218, Feb. 2024, doi: 10.1016/j.autcon.2023.105218.

A. Aboah, B. Wang, J. Bagian, and U. Nowacki, "Vision-based anomaly detection in construction: A systematic review," Advanced Engineering Informatics, vol. 55, p. 101865, Jan. 2023, doi: 10.1016/j.aei.2022.101865.

H. Hendri, L. N. Rani, S. Enggari, A. Ramadhanu, and F. Hadi, "Computer vision-based non-destructive evaluation of concrete casting using NIW and texture fusion," in Proc. 2025 Int. Conf. on Informatics, Multimedia, Cyber and Information System (ICIMCIS), Padang, Indonesia, 2025, pp. 1–6, doi: 10.1109/ICIMCIS68501.2025.11327055.

S. Dong, Z. Yang, and F. Wang, "Performance evaluation of crack detection methods in concrete structures: Handcrafted features vs. deep learning," Structures, vol. 59, p. 105742, Jan. 2024, doi: 10.1016/j.istruc.2023.105742.

J. Chen, L. Wan, and R. Zhang, "Semantic segmentation of concrete surface defects using an improved U-Net with attention mechanism," Journal of Computing in Civil Engineering, vol. 37, no. 4, p. 04023012, Jul. 2023, doi: 10.1061/JCCEO4.CPENG-5123.

M. Ellenberg, L. Kontrus, I. L. Gee, and A. Emin, "Robustness of deep learning models for civil infrastructure inspection under varying environmental conditions," Automation in Construction, vol. 148, p. 104763, Apr. 2023, doi: 10.1016/j.autcon.2023.104763.

P. Arafin, A. H. M. M. Billah, and A. Issa, "Deep learning-based concrete defects classification and detection using semantic segmentation," Structural Health Monitoring, vol. 23, no. 1, pp. 383–409, Jan. 2024, doi: 10.1177/14759217231170000.

Q. Song et al., "Two-stage framework with improved U-Net based on self-supervised contrastive learning for pavement crack segmentation," Expert Systems with Applications, vol. 238, p. 122406, Mar. 2024, doi: 10.1016/j.eswa.2023.122406.

X. Dong, Y. Liu, and J. Dai, "Concrete surface crack detection algorithm based on improved YOLOv8," Sensors, vol. 24, no. 16, p. 5252, Aug. 2024, doi: 10.3390/s24165252.

H. Jiang, J. Tan, and J. Zhang, "Comparative study of deep learning and traditional machine learning for concrete damage recognition," Archives of Computational Methods in Engineering, vol. 30, no. 6, pp. 3845–3867, Sep. 2023, doi: 10.1007/s11831-023-09923-4.

M. Pan, L. Zhang, and R. J. Y. Loo, "Digital twin-enabled real-time quality monitoring for precast concrete construction," Journal of Management in Engineering, vol. 40, no. 1, p. 04023055, Jan. 2024, doi: 10.1061/JMENEA.MEENG-5621.

S. Liu, W. Sun, and X. Guo, "Intelligent inspection of concrete structures: A review on algorithm generalization and real-world application," Measurement, vol. 226, p. 114152, Feb. 2024, doi: 10.1016/j.measurement.2023.114152.

Y. Zhao, H. Li, and J. Zhang, "A three-step computer vision-based framework for concrete crack detection and dimensions identification," Buildings, vol. 14, no. 8, p. 2360, Aug. 2024, doi: 10.3390/buildings14082360.

D. A. Klyuev, A. V. Proskuryakov, and A. A. Shcherbakov, "Computer vision method for automatic detection of microstructure defects of concrete," Sensors, vol. 24, no. 13, p. 4373, Jul. 2024, doi: 10.3390/s24134373.

Z. Wu, S. Wang, and J. Liu, "Deep learning-based surface crack detection: A comprehensive review and future perspectives," Measurement, vol. 220, p. 113339, Oct. 2023, doi: 10.1016/j.measurement.2023.113339.

H. T. Thai, "Deep learning for crack detection in concrete: From classification to segmentation," Automation in Construction, vol. 156, p. 105126, Dec. 2023, doi: 10.1016/j.autcon.2023.105126.

L. Zhang, X. Yang, and Y. Zhang, "Pixel-level quantification of concrete surface defects using deep semantic segmentation networks," Journal of Building Engineering, vol. 82, p. 108256, Apr. 2024, doi: 10.1016/j.jobe.2023.108256.

W. Zhang, Z. Zhang, and D. Qi, "Crack detection using deep learning: A review of U-Net-based methods," Archives of Computational Methods in Engineering, vol. 30, no. 7, pp. 4283–4305, Nov. 2023, doi: 10.1007/s11831-023-09941-2.

X. Wang, Y. Su, and S. Zhang, "Attention-guided U-Net with multi-scale feature fusion for concrete surface crack detection," Construction and Building Materials, vol. 411, p. 134267, Jan. 2024, doi: 10.1016/j.conbuildmat.2023.134267.

L. G. Moffatt, M. A. N. Hassan, and O. H. A. Ibrahim, "Benchmarking deep learning segmentation models for civil infrastructure defect assessment on small datasets," Journal of Civil Structural Health Monitoring, vol. 14, no. 1, pp. 121–139, Feb. 2024, doi: 10.1007/s13349-023-00728-x.

A. Ramadhanu, H. Hendri, Mardison, L. Navia Rani, S. Enggari, and M. Reza Putra, “Three Layer Median Filter Method For Identifying Concrete Strength Levels Based On Concrete Images”, IJSECS, vol. 10, no. 2, pp. 159–172, Oct. 2025, doi: 10.15282/ijsecs.10.2.2024.13.0131.

Y. Liu, S. Wang, and H. Zhao, "Efficient deep learning models for real-time infrastructure inspection," Computer-Aided Civil and Infrastructure Engineering, vol. 38, no. 5, pp. 620–635, May 2023, doi: 10.1111/mice.12930.

C. Chen, Z. Li, and J. Liu, "Data augmentation strategies for deep learning-based concrete crack detection under complex illumination," Structural Health Monitoring, vol. 21, no. 4, pp. 1532–1548, Jul. 2022, doi: 10.1177/14759217211030910.

H. Hendri, L. N. Rani, S. Enggari, A. Ramadhanu and F. Hadi, "Computer Vision-Based Non-Destructive Evaluation of Concrete Casting Using NIW and Texture Fusion," 2025 International Conference on Informatics, Multimedia, Cyber and Information System (ICIMCIS), Jakarta, Indonesia, 2025, pp. 26-31, doi: 10.1109/ICIMCIS68501.2025.11327055.

A. Ramadhanu, H. Hendri, M. A. Majid, S. Enggari, S. Andini, and R. Hidayat, "Optimization of Shape, Texture, and Color Extraction Methods in Concrete Strength Detection," JOIV: International Journal on Informatics Visualization, vol. 9, no. 6, pp. 2263–2271, Nov. 2025, doi: 10.62527/joiv.9.6.4164.

M. F. Hossain, A. H. M. M. Billah, and A. Issa, "Benchmarking deep learning models for automated visual inspection of civil infrastructure," Expert Systems with Applications, vol. 213, p. 118835, Mar. 2023, doi: 10.1016/j.eswa.2022.118835.

H. Hendri, Yuhandri and A. Ramadhanu, "GoogLeNet-Based Deep Learning Framework for Underwater Microplastic Classification in Marine Environments," 2025 International Conference on Informatics, Multimedia, Cyber and Information System (ICIMCIS), Jakarta, Indonesia, 2025, pp. 44-49, doi: 10.1109/ICIMCIS68501.2025.11327223.

A. Ramadhanu, Mardison, H. Hendri and F. Hadi, "Organic Fertilizer Content Detection Based on Image Segmentation and Texture Analysis," 2025 International Conference on Informatics, Multimedia, Cyber and Information System (ICIMCIS), Jakarta, Indonesia, 2025, pp. 50-55, doi: 10.1109/ICIMCIS68501.2025.11327142.

X. Dong, Y. Liu, and J. Dai, "A highly efficient semantic segmentation network for automated concrete surface crack and defect detection," Automation in Construction, vol. 160, p. 105298, Apr. 2024, doi: 10.1016/j.autcon.2024.105298.

J. Li, J. Tan, and H. Jiang, "Deep learning-based visual inspection for civil infrastructure: Addressing gradient vanishing and computational efficiency," Structural Health Monitoring, vol. 22, no. 5, pp. 2931–2950, Sep. 2023, doi: 10.1177/14759217221146241.

Q. Song et al., "Feature fusion mechanisms in U-shaped neural networks for pavement distress segmentation: A comparative study," Measurement, vol. 221, p. 113539, Nov. 2023, doi: 10.1016/j.measurement.2023.113539.

L. Zhang, X. Yang, and Y. Zhang, "Evaluation of traditional color-space clustering methods versus deep learning for concrete defect segmentation," Construction and Building Materials, vol. 368, p. 130456, Mar. 2023, doi: 10.1016/j.conbuildmat.2023.130456.

S. Dong, Z. Yang, and F. Wang, "Machine learning approaches utilizing Gray-Level Co-occurrence Matrix (GLCM) for structural health monitoring," Journal of Building Engineering, vol. 57, p. 104868, Oct. 2022, doi: 10.1016/j.jobe.2022.104868.

A. K. Sharma, Y. Chen, and M. R. Hosseini, "Beyond pixel accuracy: A critical review of evaluation metrics for deep learning-based structural defect segmentation," Automation in Construction, vol. 165, p. 105520, Sep. 2024, doi: 10.1016/j.autcon.2024.105520.

T. Nguyen, J. Park, and S. Kim, "Comparative analysis of semantic segmentation loss functions and metrics for imbalanced concrete crack datasets," Computer-Aided Civil and Infrastructure Engineering, vol. 40, no. 2, pp. 215–234, Feb. 2025, doi: 10.1111/mice.13115.

R. Martinez, L. Wang, and Z. Sun, "Precision-recall trade-offs in automated visual inspection of civil infrastructure: A reliability perspective," Structural Health Monitoring, vol. 23, no. 1, pp. 89–104, Jan. 2024, doi: 10.1177/14759217231189432.

S. Kim, J. Park, and Y. Lee, "Performance evaluation of deep learning architectures versus traditional computer vision for automated concrete defect quantification," Automation in Construction, vol. 168, p. 105612, Jan. 2025, doi: 10.1016/j.autcon.2024.105612.

H. Zhang, T. Nguyen, and C. Wang, "Addressing the precision-recall trade-off in structural anomaly detection: A critical review of over-segmentation issues," Measurement, vol. 224, p. 113789, Mar. 2024, doi: 10.1016/j.measurement.2023.113789.

F. Al-Hussein and M. R. Hosseini, "Dice coefficient optimization in highly imbalanced semantic segmentation tasks for civil infrastructure," Computer-Aided Civil and Infrastructure Engineering, vol. 38, no. 12, pp. 1588–1605, Dec. 2023, doi: 10.1111/mice.13045.

A. Z. Abidin, M. S. Hossain, and Y. T. Lee, "Overcoming over-segmentation and photometric artifacts in traditional concrete inspection: A semantic approach," Journal of Civil Structural Health Monitoring, vol. 15, no. 2, pp. 345–362, Feb. 2025, doi: 10.1007/s13349-024-00812-z.

Y. Zhang, L. Wang, and Z. Sun, "Hierarchical feature extraction for robust concrete surface anomaly detection under complex illumination," Structural Health Monitoring, vol. 23, no. 4, pp. 2105–2122, Jul. 2024, doi: 10.1177/14759217231201556.

K. T. Nguyen, R. Martinez, and C. Wang, "The role of skip connections in mapping irregular boundaries of structural degradation: A deep learning perspective," Automation in Construction, vol. 165, p. 105688, Sep. 2024, doi: 10.1016/j.autcon.2024.105688.

C. Chen et al., "From pixels to structural safety: Interpreting deep learning segmentation masks for concrete defect quantification," Journal of Building Engineering, vol. 85, p. 108699, May 2024, doi: 10.1016/j.jobe.2024.108699.

M. A. Qureshi, Y. Chen, and S. H. Kim, "Deep learning-based crack detection and prediction for structural health monitoring: Addressing robustness in noisy environments," Journal of Civil Structural Health Monitoring, vol. 15, no. 1, pp. 112–128, Jan. 2025, doi: 10.1007/s13349-024-00855-2.

T. R. Silva, J. Park, and L. Wang, "Segmentation of spatial crack structures in concrete by deep learning: Enabling cross-domain generalization," Journal of Soft Computing in Civil Engineering, vol. 9, no. 2, pp. 45–61, May 2025, doi: 10.22115/SCCE.2025.199726.

H. Zhang et al., "An innovative Dense ResU-Net architecture for advanced anomaly detection in concrete structures: A focus on scalability and generalization," IEEE Open Journal of the Computer Society, vol. 5, pp. 215–229, Mar. 2024, doi: 10.1109/OJCS.2024.3372020.




DOI: https://doi.org/10.47738/jads.v7i2.1298

Refbacks

  • There are currently no refbacks.


Barcode

Journal of Applied Data Sciences

ISSN:2723-6471 (Online)
Publisher:Bright Publisher
Website:http://bright-journal.org/JADS
Email:taqwa@amikompurwokerto.ac.id (principal contact)
  support@bright-journal.org (technical issues)

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0