A Generalizable Deep Learning Approach to Classifying Ground Radar Images

Abstract

Ensuring the safety of underground structures depends on the rapid and reliable detection of potential ground defects. Ground Penetrating Radar (GPR) is one of the widely used non-destructive testing methods for this purpose. However, traditional GPR data analysis has limitations in terms of both time and accuracy due to operator dependency and high workload. In this study proposes a generalizable deep learning approach for the automatic classification of GPR images. Within the scope of the study, EfficientNet, MobileNetV2, and VGG16 architectures were comparatively evaluated using two independent GPR datasets with different structural features and reflection characteristics. To objectively examine the generalization capabilities of the models, 5-fold cross-validation was applied and their performance was analyzed using Accuracy, Precision, Recall, and F1-Score metrics. The experimental results demonstrate that deep learning architectures can accurately distinguish structural anomalies in GPR data. MobileNetV2 model showed the highest performance, achieving 100% accuracy on the Underground Utilities Dataset (UUD) and 97.46% accuracy on the TIGPR dataset. The findings demonstrate that GPR images can be successfully classified using deep learning architectures, and this approach has strong potential in applications such as infrastructure health monitoring, tunnel lining assessment, and subsurface defect detection.

Keywords

• Ground penetrating radar (GPR)
• Subsurface anomaly detection
• Deep learning
• Convolutional neural networks (CNN)

Yer Radarı Görüntülerinin Sınıflandırılmasına Yönelik Genelleştirilebilir Bir Derin Öğrenme Yaklaşımı

Abstract

Yeraltı yapılarının güvenliğinin sağlanması, potansiyel zemin kusurlarının hızlı ve güvenilir bir şekilde tespit edilmesine bağlıdır. Yer Radarı (GPR), bu amaçla yaygın olarak kullanılan tahribatsız test yöntemlerinden biridir. Bununla birlikte, geleneksel GPR veri analizi, operatör bağımlılığı ve yüksek iş yükü nedeniyle hem zaman hem de doğruluk açısından sınırlamalara sahiptir. Bu çalışmada, GPR görüntülerinin otomatik sınıflandırılması için genelleştirilebilir bir derin öğrenme yaklaşımı önerilmektedir. Çalışma kapsamında, EfficientNet, MobileNetV2 ve VGG16 mimarileri, farklı yapısal özelliklere ve yansıma karakteristiklerine sahip iki bağımsız GPR veri seti kullanılarak karşılaştırmalı olarak değerlendirilmiştir. Modellerin genelleştirme yeteneklerini objektif olarak incelemek için 5 katlı çapraz doğrulama uygulanmış ve performansları Doğruluk, Hassasiyet, Geri Çağırma ve F1-Skor metrikleri kullanılarak analiz edilmiştir. Deneysel sonuçlar, derin öğrenme mimarilerinin GPR verilerindeki yapısal anomalileri yüksek doğrulukla ayırt edebildiğini göstermektedir. MobileNetV2 modeli, Yeraltı Altyapı Veri Kümesi (UUD) üzerinde %100 doğruluk ve TIGPR veri kümesi üzerinde %97,46 doğruluk elde ederek en yüksek performansı göstermiştir. Bulgular, GPR görüntülerinin derin öğrenme mimarileri kullanılarak başarılı bir şekilde sınıflandırılabileceğini ve bu yaklaşımın altyapı sağlığı izleme, tünel kaplama değerlendirmesi ve yer altı kusur tespiti gibi uygulamalarda güçlü bir potansiyele sahip olduğunu göstermektedir.

References

1. Alani, A. M., & Tosti, F. (2018). GPR applications in structural detailing of a major tunnel using different frequency antenna systems. Construction and Building Materials, 158, 1111–1122. https://doi.org/10.1016/j.conbuildmat.2017.09.100
2. İpek, S., Işık, N., & Halifeoğlu, F. M. (2023). DETERMINATION OF GROUND-BASED STRUCTURAL PROBLEMS IN THE HISTORICAL FOUR-LEGGED MINARET WITH GROUND PENETRATION RADAR. Turkish Journal of Nature and Science, 12(2), 119-131. https://doi.org/10.46810/tdfd.1224164
3. Geng, Q., Ye, Y., & Wang, X. (2022). Identifying void defects behind Tunnel composite lining based on transient electromagnetic radar method. NDT & E International, 125, 102562. https://doi.org/10.1016/j.ndteint.2021.102562
4. Rasol, M., Elseicy, A., Solla, M., Celaya, M., & Schmidt, F. (2024). Role of intelligent data analysis to enhance GPR data interoperability: road transports. In Interpretable Machine Learning for the Analysis, Design, Assessment, and Informed Decision Making for Civil Infrastructure (pp. 159–184). Elsevier. https://doi.org/10.1016/b978-0-12-824073-1.00013-7
5. Ponti, F., Barbuto, F., Di Gregorio, P. P., Frezza, F., Mangini, F., Parisi, R., Simeoni, P., & Troiano, M. (2021). GPR radargrams analysis through machine learning approach. Journal of Electromagnetic Waves and Applications, 35(12), 1678–1686. https://doi.org/10.1080/09205071.2021.1906329
6. Liu, H., Yue, Y., Liu, C., Spencer, B. F., Jr, & Cui, J. (2023). Automatic recognition and localization of underground pipelines in GPR B-scans using a deep learning model. Tunnelling and Underground Space Technology, 134, 104861. https://doi.org/10.1016/j.tust.2022.104861
7. Erdaş, S., Akkaya, A. E., & Aydın, A. A. (2025). Machine Learning Based Hybrid DDoS Attack Prediction. European Journal of Technique (EJT), 15(2), 231-241. https://doi.org/10.36222/ejt.1670798
8. Jin, Y., & Duan, Y. (2020). Wavelet Scattering Network-Based Machine Learning for Ground Penetrating Radar Imaging: Application in Pipeline Identification. Remote Sensing, 12(21), 3655. https://doi.org/10.3390/rs12213655

9. Pham, M.-T., & Lefevre, S. (2018). Buried Object Detection from B-Scan Ground Penetrating Radar Data Using Faster-RCNN. In IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium (pp. 6804–6807). IEEE. IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium. https://doi.org/10.1109/igarss.2018.8517683
10. Qin, H., Zhang, D., Tang, Y., & Wang, Y. (2021). Automatic recognition of tunnel lining elements from GPR images using deep convolutional networks with data augmentation. Automation in Construction, 130, 103830. https://doi.org/10.1016/j.autcon.2021.103830
11. Yue, Y., Liu, H., Lin, C., Meng, X., Liu, C., Zhang, X., Cui, J., & Du, Y. (2024). Automatic recognition of defects behind railway tunnel linings in GPR images using transfer learning. Measurement, 224, 113903. https://doi.org/10.1016/j.measurement.2023.113903
12. Wang, Y., Qin, H., Zhang, D., Wu, T., & Pan, S. (2024). Domain Adaption YOLO Network to Enhance Target Detection in GPR Images. IEEE Transactions on Geoscience and Remote Sensing, 62, 1–15. https://doi.org/10.1109/tgrs.2024.3505946
13. Yang, Y., Huang, L., Zhang, Z., Zhang, J., & Zhao, G. (2024). CycleGAN-Based Data Augmentation for Subgrade Disease Detection in GPR Images with YOLOv5. Electronics, 13(5), 830. https://doi.org/10.3390/electronics13050830
14. Liu, C., Yao, Y., Li, J., Qian, J., & Liu, L. (2024). Research on lightweight GPR road surface disease image recognition and data expansion algorithm based on YOLO and GAN. Case Studies in Construction Materials, 20, e02779. https://doi.org/10.1016/j.cscm.2023.e02779
15. Zhang, B., Wang, X., Ding, L., Zang, Q., Cong, B., Cai, H., Liu, T., & Zhong, Y. (2024). Identification and 3D reconstruction of semi-rigid base loose distress from GPR B-scan using Generative Adversarial Network. Construction and Building Materials, 455, 139081. https://doi.org/10.1016/j.conbuildmat.2024.139081
16. Luo, J., Lei, W., Hou, F., Wang, C., Ren, Q., Zhang, S., Luo, S., Wang, Y., & Xu, L. (2021). GPR B-Scan Image Denoising via Multi-Scale Convolutional Autoencoder with Data Augmentation. Electronics, 10(11), 1269. https://doi.org/10.3390/electronics10111269
17. Wang, Y., Qin, H., Tang, Y., Zhang, D., Yang, D., Qu, C., & Geng, T. (2022). RCE-GAN: A Rebar Clutter Elimination Network to Improve Tunnel Lining Void Detection from GPR Images. Remote Sensing, 14(2), 251. https://doi.org/10.3390/rs14020251
18. Feng, D., Wang, X., Wang, X., Ding, S., & Zhang, H. (2021). Deep Convolutional Denoising Autoencoders with Network Structure Optimization for the High-Fidelity Attenuation of Random GPR Noise. Remote Sensing, 13(9), 1761. https://doi.org/10.3390/rs13091761
19. Gao, K., Donahue, C., Henderson, B. G., & Modrak, R. T. (2022). Deep-learning-guided high-resolution subsurface reflectivity imaging with application to ground-penetrating radar data. Geophysical Journal International, 233(1), 448–471. https://doi.org/10.1093/gji/ggac468
20. Wickramanayake, S., Thiyagarajan, K., & Kodagoda, S. (2022). Deep Learned Ground Penetrating Radar Subsurface Features for Robot Localization. In 2022 IEEE Sensors (pp. 1–4). IEEE. 2022 IEEE Sensors. https://doi.org/10.1109/sensors52175.2022.9967350
21. Kang, M.-S., Kim, N., Im, S. B., Lee, J.-J., & An, Y.-K. (2019). 3D GPR Image-based UcNet for Enhancing Underground Cavity Detectability. Remote Sensing, 11(21), 2545. https://doi.org/10.3390/rs11212545
22. Kim, N., Kim, K., An, Y.-K., Lee, H.-J., & Lee, J.-J. (2018). Deep learning-based underground object detection for urban road pavement. International Journal of Pavement Engineering, 21(13), 1638–1650. https://doi.org/10.1080/10298436.2018.1559317
23. Su, Y., Wang, J., Li, D., Wang, X., Hu, L., Yao, Y., & Kang, Y. (2023). End-to-end deep learning model for underground utilities localization using GPR. Automation in Construction, 149, 104776. https://doi.org/10.1016/j.autcon.2023.104776
24. Zhang, X., Han, L., Robinson, M., & Gallagher, A. (2021). A Gans-Based Deep Learning Framework for Automatic Subsurface Object Recognition From Ground Penetrating Radar Data. IEEE Access, 9, 39009–39018. https://doi.org/10.1109/access.2021.3064205
25. Solla, M., Pérez-Gracia, V., & Fontul, S. (2021). A Review of GPR Application on Transport Infrastructures: Troubleshooting and Best Practices. Remote Sensing, 13(4), 672. https://doi.org/10.3390/rs13040672
26. Dérobert, X., & Pajewski, L. (2018). TU1208 Open Database of Radargrams: The Dataset of the IFSTTAR Geophysical Test Site. Remote Sensing, 10(4), 530. https://doi.org/10.3390/rs10040530
27. Alani, A. M., Tosti, F., Ciampoli, L. B., Gagliardi, V., & Benedetto, A. (2020). An integrated investigative approach in health monitoring of masonry arch bridges using GPR and InSAR technologies. NDT & E International, 115, 102288. https://doi.org/10.1016/j.ndteint.2020.102288
28. El Karakhi, S., Reineix, A., & Guiffaut, C. (2024). Multi-label classification with deep learning techniques applied to the B-Scan images of GPR. Comptes Rendus. Physique, 25(S1), 109–124. https://doi.org/10.5802/crphys.193
29. Liu, Z., Cui, B., & Gu, X. (2025). TIGPR: A multi-view ground penetrating radar detection data for damage assessment of transportation infrastructure. Data in Brief, 60, 111665. https://doi.org/10.1016/j.dib.2025.111665
30. Mojahid, A., Ouai, D. E., Amraoui, K. E., El-Hami, K., & Aitbenamer, H. (2025). Intelligent recognition of subsurface utilities and voids: A ground penetrating radar dataset for deep learning applications. Data in Brief, 59, 111338. https://doi.org/10.1016/j.dib.2025.111338
31. Rajwal, S., & Aggarwal, S. (2023). Convolutional Neural Network-Based EEG Signal Analysis: A Systematic Review. Archives of Computational Methods in Engineering, 30(6), 3585–3615. https://doi.org/10.1007/s11831-023-09920-1
32. Alçin, Z. M., & Aslan, M. (2026). Apricot Disease Detection with Convolutional Neural Network. Türk Doğa ve Fen Dergisi, 15(1), 31–42. https://doi.org/10.46810/tdfd.1774549
33. Agarwal, N., Bullins, B., Chen, X., Hazan, E., Singh, K., Zhang, C. & Zhang, Y.. (2019). Efficient Full-Matrix Adaptive Regularization. Proceedings of the 36th International Conference on Machine Learning, in Proceedings of Machine Learning Researc 97:102-110 Available from https://proceedings.mlr.press/v97/agarwal19b.html.
34. Sandler, M., Howard, A., Zhu, M., Zhmoginov, A., & Chen, L.-C. (2018). MobileNetV2: Inverted Residuals and Linear Bottlenecks. In 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 4510–4520). IEEE. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). https://doi.org/10.1109/cvpr.2018.00474
35. Simonyan, K., & Zisserman, A. (2014). Very Deep Convolutional Networks for Large-Scale Image Recognition (Version 6). arXiv. https://doi.org/10.48550/ARXIV.1409.1556
36. Muschelli, J., III. (2019). ROC and AUC with a Binary Predictor: a Potentially Misleading Metric. Journal of Classification, 37(3), 696–708. https://doi.org/10.1007/s00357-019-09345-1