지구물리와물리탐사 (Geophysics and Geophysical Exploration)

  • 제26권4호
  • /
  • Pages.211-228
  • /
  • 2023
  • /
  • 1229-1064(pISSN)
  • /
  • 2384-051X(eISSN)
한국지구물리·물리탐사학회 (Korean Society of Earth and Exploration Geophysicists)
권호 표지
DOI QR코드

DOI QR Code

데이터 증강 및 앙상블 기법을 이용한 딥러닝 기반 GPR 공동 탐지 모델 성능 향상 연구

Improving the Performance of Deep-Learning-Based Ground-Penetrating Radar Cavity Detection Model using Data Augmentation and Ensemble Techniques

  • 최용욱 (전남대학교 에너지자원공학과) ;
  • 서상진 (한국농어촌공사 농어촌연구원) ;
  • 장한길로 (보민글로벌(주)) ;
  • 윤대웅 (전남대학교 에너지자원공학과)
  • Yonguk Choi (Dept. Energy & Resources Engineering, Chonnam National University) ;
  • Sangjin Seo (Rural Research Institute, Korea Rural Community Corporation) ;
  • Hangilro Jang (Bomin Global Co.) ;
  • Daeung Yoon (Dept. Energy & Resources Engineering, Chonnam National University)
  • 투고 : 2023.10.06
  • 심사 : 2023.11.22
  • 발행 : 2023.11.30
PDF 다운로드
⟨ 이전 논문 다음 논문 ⟩

초록

방조제의 모니터링에는 지구물리학적 비파괴 검사인 GPR (Ground Penetrating Radar) 탐사가 주로 이용된다. GPR 반응은 상황에 따라 복잡한 양상을 보이므로 자료의 처리와 해석은 전문가의 주관적 판단에 의존하며, 이는 오 탐지의 가능성을 불러옴과 동시에 시간이 오래 걸린다는 단점이 있다. 따라서 딥 러닝을 이용하여 GPR 탐사자료의 공동을 탐지하는 다양한 연구들이 수행되고 있다. 딥 러닝 기반 방법은 데이터 기반 방법으로써 풍부한 자료가 필요하나 GPR 탐사의 경우 비용 등의 이유로 학습에 이용할 현장 자료가 부족하다. 따라서 본 논문에서는 데이터 증강 전략을 이용하여 딥 러닝 기반 방조제 GPR 탐사자료 공동 탐지 모델을 개발하였다. 다년간 동일한 방조제에서 탐사 자료를 사용하여 데이터 세트를 구축하였으며, 컴퓨터 비전 분야의 객체 탐지 모델 중 YOLO (You Look Only Once) 모델을 이용하였다. 데이터 증강 전략을 비교 및 분석함으로써 최적의 데이터 증강 전략을 도출하였고, 초기 모델 개발 후 앵커 박스 클러스터링, 전이 학습, 자체 앙상블, 모델 앙상블 기법을 단계적으로 적용하여 최종 모델 도출 후 성능을 평가하였다.

Ground-penetrating radar (GPR) surveys are commonly used to monitor embankments, which is a nondestructive geophysical method. The results of GPR surveys can be complex, depending on the situation, and data processing and interpretation are subject to expert experiences, potentially resulting in false detection. Additionally, this process is time-intensive. Consequently, various studies have been undertaken to detect cavities in GPR survey data using deep learning methods. Deep-learning-based approaches require abundant data for training, but GPR field survey data are often scarce due to cost and other factors constaining field studies. Therefore, in this study, a deep- learning-based model was developed for embankment GPR survey cavity detection using data augmentation strategies. A dataset was constructed by collecting survey data over several years from the same embankment. A you look only once (YOLO) model, commonly used in computer vision for object detection, was employed for this purpose. By comparing and analyzing various strategies, the optimal data augmentation approach was determined. After initial model development, a stepwise process was employed, including box clustering, transfer learning, self-ensemble, and model ensemble techniques, to enhance the final model performance. The model performance was evaluated, with the results demonstrating its effectiveness in detecting cavities in embankment GPR survey data.

키워드

과제정보

본 연구는 한국농어촌공사 농어촌연구원의 학술연구용역인' 딥러닝을 이용한 GPR 모니터링 자료 공동 반응 분석 및 방조제 현장적용 연구'사업의 지원으로 수행되었으며, 이에 감사드립니다.

참고문헌

  1. Al-Nuaimy, W., Huang, Y., Nakhkash, M., Fang, M. T. C., Nguyen, V. T., and Eriksen, A., 2000, Automatic detection of buried utilities and solid objects with GPR using neural networks and pattern recognition, J. Appl. Geophy., 43(2-4), 157-165. doi: 10.1016/S0926-9851(99)00055-5
  2. Bochkovskiy, A., Wang, C. Y., and Liao, H. Y. M., 2020, Yolov4: Optimal speed and accuracy of object detection, arXiv preprint arXiv:2004.10934. doi: 10.48550/arXiv.2004.10934
  3. Chae, J., Ko, H. Y., Lee, B. G., and Kim, N., 2019, A study on the pipe position estimation in GPR images using deep learning based convolutional neural network, Journal of Internet Computing and Services, 20(4), 39-46 (in Korean with English abstract). doi: 10.7472/jksii.2019.20.4.39
  4. Choi, B. H., Pyun, S. J., Choi, W. C., Cho, C, H., and Yoon, J. S., 2022, Deep-learning-based GPR Data Interpretation Technique for Detecting Cavities in Urban Roads. Geophys. and Geophys. Explor., 25(4), 189-200. doi: 10.7582/GGE.2022.25.4.189
  5. Choi, Y. G., Seo, S. J., and Suh, J. H., 2001, Dipole Antennas and Radiation Patterns in the Three-Dimensional GPR Modeling, Geophys. and Geophys. Explor., 4(2), 45-54 (in Korean with English abstract). http://koreascience.or.kr/article/JAKO200107921803414.page 107921803414.page
  6. Dalal, N., and Triggs, B., 2005, Histograms of oriented gradients for human detection. CVPR 2005, Proceedings of the IEEE computer society conference on computer vision and pattern recognition, p. 886-893. doi: 10.1109/CVPR.2005.177
  7. Deng, J., Dong, W., Socher, R., Li, L. J., Li, K., and Fei-Fei, L., 2009, Imagenet: A large-scale hierarchical image database, CVPR 2009, Proceedings of the IEEE conference on computer vision and pattern recognition, p. 248-255. doi: 10.1109/CVPR.2009.5206848
  8. El-Mahallawy, M. S., and Hashim, M., 2013, Material classification of underground utilities from GPR images using DCTbased SVM approach, IEEE Geosci. Remote Sens. Lett., 10(6), 1542-1546. doi: 10.1109/LGRS.2013.2261796
  9. Feng, D., Wang, X., and Zhang, B., 2018, Specific evaluation of tunnel lining multi-defects by all-refined GPR simulation method using hybrid algorithm of FETD and FDTD. Construction and Building Materials, 185, 220-229. doi: 10.1016/j.conbuildmat.2018.07.039
  10. Gao, R., Zhu, H., Liao, Q., Qu, B., Hu, L., and Wang, H., 2022, Detection of coal fire by deep learning using ground penetrating radar, Measurement, 201, 111585. doi: 10.1016/j.measurement.2022.111585
  11. Girshick, R., Donahue, J., Darrell, T., and Malik, J., 2014, Rich feature hierarchies for accurate object detection and semantic segmentation. CVPR 2014, Proceedings of the IEEE conference on computer vision and pattern recognition, p. 580-587. doi: 10.1109/CVPR.2014.81
  12. Girshick, R., 2015, Fast r-cnn. ICCV 2015, Proceedings of the IEEE International Conference on Computer Vision, p. 1440-1448. doi: 10.1109/ICCV.2015.169
  13. Jeon, S., Kim, D., and Jung, H., 2021, YOLO-based lane detection system, Journal of the Korea Institute of Information and Communication Engineering, 25(3), 464-470 (in Korean with English abstract). doi: 10.6109/jkiice.2021.25.3.464
  14. Kim, N., Kim, S., An, Y. K., and Lee, J. J., 2019, Triplanar imaging of 3-D GPR data for deep-learning-based underground object detection, IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., 12(11), 4446-4456. doi: 10.1109/JSTARS.2019.2953505
  15. Kim, N., Kim, S., An, Y. K., and Lee, J. J., 2021, A novel 3D GPR image arrangement for deep learning-based underground object classification, International Journal of Pavement Engineering, 22(6), 740-751. doi: 10.1080/10298436.2019.1645846
  16. Kong, H. J., Kim, S. D., Kim, M., and Han, S. H., 2013, A Preprocessing Method for Ground-Penetrating-Radar based Land-mine Detection System, Journal of the Institute of Electronics and Information Engineers, 50(4), 171-181 (in Korean with English abstract). https://www.kci.go.kr/kciportal/ci/sereArticleSearch/ciSereArtiView.kci?sereArticleSearchBean.artiId=ART001761316 https://doi.org/10.5573/ieek.2013.50.4.171
  17. 한국농어촌공사, 2023, 농업생산기반정비통계조사 방조제 현황, https://kosis.kr/statHtml/statHtml.do?orgId=311&tblId=DT_311001_016&lang_mode=ko&vw_cd=MT_ZTITLE&list_id=311_31101_1_1&conn_path=I4 (November 10, 2023 Accessed)
  18. Lestari, D. P., Kosasih, R., Handhika, T., Sari, I., and Fahrurozi, A., 2019, Fire hotspots detection system on CCTV videos using you only look once (YOLO) method and tiny YOLO model for high buildings evacuation, IC2IE 2019, 2019 2nd International Conference of Computer and Informatics Engineering (IC2IE): Proceedings : "Artificial Intelligence Roles in Industrial Revolution 4.0, p. 87-92. doi: 10.1109/IC2IE47452.2019.8940842
  19. Li, Y., Zhao, Z., Luo, Y., and Qiu, Z., 2020, Real-time pattern-recognition of GPR images with YOLO V3 implemented by TensorFlow, Sensors, 20(22), 6476. doi: 10.3390/s20226476
  20. Lim, B., Son, S., Kim, H., Nah, S., and Lee, K. M., 2017, Enhanced deep residual networks for single image super-resolution, CVPRW 2017, Proceedings of the IEEE conference on computer vision and pattern recognition workshops, p. 136-144. doi: 10.1109/CVPRW.2017.151
  21. Lin, T. Y., Maire, M., Belongie, S., Hays, J., Perona, P., Ramanan, D., Dollar, P., and Zitnick, C. L., 2014, Microsoft coco: Common objects in context, ECCV 2014, Computer Vision - ECCV 2014: 13th European Conference proceedings, p. 740-755. doi: 10.48550/arXiv.1405.0312
  22. Liu, Z., Wu, W., Gu, X., Li, S., Wang, L., and Zhang, T., 2021, Application of combining YOLO models and 3D GPR images in road detection and maintenance, Remote Sens., 13(6), 1081. doi: 10.3390/rs13061081
  23. Liu, Z., Gu, X., Wu, W., Zou, X., Dong, Q., and Wang, L., 2022, GPR-based detection of internal cracks in asphalt pavement: A combination method of DeepAugment data and object detection, Measurement, 197, 111281. doi: 10.1016/j.measurement.2022.111281
  24. 국토지리정보원, 2020, 대한민국국가지도집 II, 2, 국토지리정보원, 242-245. http://nationalatlas.ngii.go.kr/pages/page_200.php
  25. Nugraha, B. T., and Su, S. F., 2017, Towards self-driving car using convolutional neural network and road lane detector, ICACOMIT 2017, Proceedings of the 2017 2nd international conference on automation, cognitive science, optics, micro electro-mechanical system, and information technology(ICACOMIT), p. 65-69. doi: 10.1109/ICACOMIT.2017.8253388
  26. Qiu, Z., Zhao, Z., Chen, S., Zeng, J., Huang, Y., and Xiang, B., 2022, Application of an improved YOLOv5 algorithm in real-time detection of foreign objects by ground penetrating radar, Remote Sens., 14(8), 1895. doi: 10.3390/rs14081895
  27. Redmon, J., Divvala, S., Girshick, R., and Farhadi, A., 2016, You only look once: Unified, real-time object detection, CVPR 2016, Proceedings of the IEEE conference on computer vision and pattern recognition, p. 779-788. doi: 10.1109/CVPR.2016.91
  28. Redmon, J., and Farhadi, A., 2017, YOLO9000: better, faster, stronger, CVPR 2017, Proceedings of the IEEE conference on computer vision and pattern recognition, p. 7263-7271. doi: 10.1109/CVPR.2017.690
  29. Redmon, J., and Farhadi, A., 2018, Yolov3: An incremental improvement, arXiv preprint arXiv:1804.02767. doi: 10.48550/arXiv.1804.02767
  30. Ren, S., He, K., Girshick, R., and Sun, J., 2015, Faster R-CNN: towards real-time object detection with region proposal networks, NIPS'15, Proceedings of the 28th International Conference on Neural Information Processing Systems, 1, p. 91-99. doi: 10.1109/TPAMI.2016.2577031
  31. Sagi, O., and Rokach, L., 2018, Ensemble learning: A survey. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery, 8(4), e1249. doi: 10.1002/widm.1249
  32. Song, S. H., Won, J. K., Choi, J. H., and Kim, J. H., 1999, Embankment and sea dike leak investigation using geophysical methods, 54th Fall Joint Academic Conference in 1999, Conference of the geological society of korea Expanded Abstracts, p. 14-14 (in korean). http://www.riss.kr/link?id=A82544214
  33. Song, S. H., Lee, K. S., Kim, J. H., and Kwon, B. D., 2000, Application of SP and Pole-pole Array Electrical Resistivity Surveys to the Seawater Leakage Problem of the Embankment, Economic and Environmental Geology, 33(5), 417-424 (in korean with English abstract). http://www.riss.kr/link?id=A100573175 100573175
  34. Song, S. H., Lee, G. S., Kim, Y. G., Seong, B. U., Kim, Y. B., and Kim, D. H., 2007, Detection of seawater inflow through embankment dike near reinforced region: Application of electric and electromagnetic surveys, Journal of Korean Society of Miniral and Energy Resources Engineers, 44(3), 235-243 (in korean with English abstract). http://www.riss.kr/link?id=A104636447 104636447
  35. Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., Erhan, D., Vanhoucke, V., and Rabinovich, A., 2015, Going deeper with convolutions, CVPR 2015, Proceedings of the IEEE conference on computer vision and pattern recognition, p. 1-9. doi: 10.1109/CVPR.2015.7298594
  36. Terven, J., and Cordova-Esparza, D, 2023, A comprehensive review of YOLO: From YOLOv1 to YOLOv8 and beyond, arXiv:2304.00501. https://doi.org/10.48550/arXiv.2304.00501
  37. Tzutalin, 2015. LabelImg Free Software: MIT License. https://github.com/tzutalin/labelImg
  38. Ultralytics/yolov5, 2023.07.26., https://github.com/ultralytics/yolov5
  39. Viola, P., and Jones, M., 2001, Rapid object detection using a boosted cascade of simple features. CVPR 2001, Proceedings of the IEEE computer society conference on computer vision and pattern recognition, p. I-I. doi: 10.1109/CVPR.2001.990517
  40. Xu, R., Lin, H., Lu, K., Cao, L., and Liu, Y., 2021, A forest fire detection system based on ensemble learning. Forests, 12(2), 217. doi: 10.3390/f12020217
  41. Xu, X., Zeng, Q., Li, D., Wu, J., Wu, X., and Shen, J., 2010, GPR detection of several common subsurface voids inside dikes and dams, Eng. Geol., 111(1-4), 31-42. doi: 10.1016/j.enggeo.2009.12.001
  42. Yee, K., 1966, Numerical solution of initial boundary value problems involving Maxwell's equations in isotropic media. IEEE Trans. Antennas Propag., 14(3), 302-307. doi: 10.1109/TAP.1966.1138693
  43. Yong, H. H., 2013, Geophysical Surveillance Techniques for the Maintenance of Sea dyke, Ph. D. thesis, Kangwon National University, p.10-11 (in korean with English abstract). http://www.riss.kr/link?id=T13263544
  44. Yong, H. H., Lee, D. Y., and Song, S. H., 2014, Safety assessment by surface geophysical survey considering cover state of the sea dike, Proceedings of the Korean Society of Agricultural Engineers Conference 2014, Expanded Abstracts, p. 110-110 (in korean). http://www.riss.kr/link?id=A105336564
  45. Yong, H. H, Lim, K. H., Lee, B. S., and Lee, G. S., 2020, Embankment safety management using multi-channel GPR exploration, Proceedings of KSEG 2020 Fall Conference, Expanded Abstracts, p. 67-68 (in korean). http://www.riss.kr/link?id=A107277262
  46. Zong, Z., Chen, C., Mi, X., Sun, W., Song, Y., Li, J., Dong, J., Huang, R., and Yang, B., 2019, A deep learning approach for urban underground objects detection from vehicle-borne ground penetrating radar data in real-time, Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., 42, 293-299. doi: 10.5194/isprs-archives-XLII-2-W16-293-2019
  47. Zoph, B., Cubuk, E. D., Ghiasi, G., Lin, T. Y., Shlens, J., and Le, Q. V, 2020, Learning data augmentation strategies for object detection, ECCV 2020, Computer Vision - ECCV 2020 16th European Conference Proceedings, 566-583. doi: 10.1007/978-3-030-58583-9_34
  48. Zou, Z., Chen, K., Shi, Z., Guo, Y., and Ye, J., 2023, Object detection in 20 years: A survey. Proceedings of the IEEE, 111(3), 257-276. doi: 10.1109/JPROC.2023.3238524