VAK Russia 1.6 UDC 550.8.053 CSCSTI 37.01 Russian Classification of Professions by Education 05.00.00 Russian Library and Bibliographic Classification 26 Russian Trade and Bibliographic Classification 63 BISAC SCI UDC 55 UDC 550.34 UDC 550.383 CSCSTI 37.15 CSCSTI 37.25 CSCSTI 37.31 CSCSTI 38.01 CSCSTI 36.00 CSCSTI 37.00 CSCSTI 38.00 CSCSTI 39.00 CSCSTI 52.00

YOLO-BASED FRACTURE DETECTION ON ELECTRICAL BOREHOLE MICROIMAGER DATA WITH A SEMI-SYNTHETIC TRAINING SET: CASE STUDY AT UTAH FORGE YOLO-Based Fracture Detection on Electrical Borehole Microimager Data with a Semi-Synthetic Training Set: Case Study at Utah FORGE

Published in Russian Journal of Earth Sciences · Volume 26, Issue 3 · ELocator: ES3015 · Rubric: ORIGINAL ARTICLES
DOI: https://doi.org/10.2205/2026es001139 · EDN: HLXPXI
Received: 06.12.2025 Accepted: 06.06.2026 Published: 30.08.2026 Language of publication: ru-RU
The assessment of natural rock fracturing from electrical borehole microimager (Fullbore Formation MicroImager, FMI) data is an important task of reservoir geomechanics, unconventional hydrocarbon exploration, and the characterization of geothermal systems. At the same time, manual fracture picking on FMI images is labor-intensive and depends substantially on the interpreter’s experience, whereas existing automated algorithms – gradient operators, the Hough transform, and convolutional neural networks tested primarily on acoustic imagers – lack robustness to noise and to the vertical artifacts typical of electrical microimagers. The aim of this work is to develop a robust methodology for automatic fracture detection on FMI images by means of deep learning. The proposed algorithm is built around a YOLOv11s neural network adapted to the detection of elongated and intersecting sinusoidal structures. To overcome the shortage of labeled data that is typical of geophysical tasks, a semi-synthetic training set was constructed: parametrically generated sinusoidal fractures were superimposed on 154 real FMI images from well 78B-32 of the Utah FORGE site, which yielded 2140 training and 310 validation images (8406 fracture instances in total); independent verification was performed on 110 real images labeled by an expert. On the validation set, the model achieved Precision = 0.85, Recall = 0.78, and mAP@50 = 0.807. On real data, the recovery of sinusoidal fracture-approximation parameters yielded Precision = 0.46 and F1 = 0.296, more than three times the values of the baseline edgedetection algorithm based on DexiNed (Precision = 0.153, F1 = 0.155). These results confirm the applicability of the semi-synthetic-training approach to FMI image interpretation tasks.
microimager, edge detection, DL, fullbore microimager, FMI, machine learning, deep learning
Funding
The author thanks the administrators of the website https://gdr.openei.org/ for providing access to the data used in this study. The author also thanks her supervisor, Academician of the Russian Academy of Sciences, Doctor of Physical and Mathematical Sciences S. A. Tikhotsky, for valuable recommendations and consultations during the research and preparation of this publication.
Text References
Text (PDF)
Read Download

1. Arkhipov N. D. Fracture detection technique based on microimager data using machine learning methods // Seismicheskie Pribory. — 2024. — Vol. 60, no. 2. — P. 53–66. — https://doi.org/10.21455/si2024.2-4 — (In Russian).

2. Dubinya N. V. and Tikhotskiy S. A. Method for the Inverse Problem Solution for Reconstruction of Stress Strain State of Rock Mass Based on Natural Fractures Data // Izvestiya, Physics of the Solid Earth. — 2022. — Vol. 58, no. 4. — P. 544–561. — https://doi.org/10.1134/s1069351322040024

3. Bochkovskiy A., Wang C.-Y. and Liao H.-Y. M. YOLOv4: Optimal Speed and Accuracy of Object Detection. — 2020. — https://doi.org/10.48550/ARXIV.2004.10934

4. Dias L. O., Bom C. R., Faria E. L., et al. Automatic detection of fractures and breakouts patterns in acoustic borehole image logs using fast-region convolutional neural networks // Journal of Petroleum Science and Engineering. — 2020. — Vol. 191. — P. 107099. — https://doi.org/10.1016/j.petrol.2020.107099

5. Diwan T., Anirudh G. and Tembhurne J. V. Object detection using YOLO: challenges, architectural successors, datasets and applications // Multimedia Tools and Applications. — 2023. — Vol. 82, no. 6. — P. 9243–9275. — https://doi.org/10.1007/s11042-022-13644-y

6. Du L., Lu X. and Li H. Automatic fracture detection from the images of electrical image logs using Mask R-CNN // Fuel. — 2023. — Vol. 351. — P. 128992. — https://doi.org/10.1016/j.fuel.2023.128992

7. Fu Zh., Lu Sh., Wang H., et al. Natural fracture occurrence model based on FMI imaging logging // ACS Omega. — 2023. — Vol. 8, no. 2. — P. 2034–2045. — https://doi.org/10.1021/acsomega.2c05262

8. Gaillot Ph., Brewer T., Pezard Ph., et al. Supplement 1: Borehole Imaging Tools – Principles and Applications // Scientific Drilling. — 2007. — No. 5. — P. 1. — https://doi.org/10.2204/iodp.sd.5.07s1.2007

9. Gonzalez R. C. and Woods R. E. Digital image processing. — 2. ed., internat. ed. — Upper Saddle River, NJ : PrenticeHall, 2002. — 793 p. — Includes bibliographical references and index. - Previous ed.:Reading, Mass..: Addison-Wesley, 1992.

10. Jones C., Simmons S. and Moore J. Geology of the Utah frontier observatory for research in geothermal energy (FORGE) enhanced geothermal system (EGS) site // Geothermics. — 2024. — Vol. 122. — P. 103054. — https://doi.org/10.1016/j.geothermics.2024.103054

11. Krizhevsky A., Sutskever I. and Hinton G. E. ImageNet Classification with Deep Convolutional Neural Networks // Advances in Neural Information Processing Systems. — Curran Associates, Inc., 2012.

12. Li X.-N., Shen J.-S., Yang W.-Ya., et al. Automatic fracture-vug identification and extraction from electric imaging logging data based on path morphology // Petroleum Science. — 2019. — Vol. 16, no. 1. — P. 58–76. — https://doi.org/10.1007/s12182-018-0282-6

13. Mazdarani A., Kadkhodaie A., Wood D. A., et al. Natural fractures characterization by integration of FMI logs, well logs and core data: a case study from the Sarvak Formation (Iran) // Journal of Petroleum Exploration and Production Technology. — 2023. — Vol. 13, no. 5. — P. 1247–1263. — https://doi.org/10.1007/s13202-023-01611-8

14. McLennan J. Utah FORGE Well 78B-32 Daily Drilling Reports and Logs. — 2021. — https://doi.org/10.15121/1814488

15. Nikolenko S. I. Synthetic data for deep learning. — Cham, Switzerland : Springer International Publ., 2021. — https://doi.org/10.1007/978-3-030-75178-4

16. Redmon J., Divvala S., Girshick R., et al. You Only Look Once: Unified, Real-Time Object Detection // 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). — IEEE, 2016. — P. 779–788. — https://doi.org/10.1109/cvpr.2016.91

17. Ronneberger O., Fischer Ph. and Brox Th. U-Net: Convolutional Networks for Biomedical Image Segmentation // Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015. — Springer International Publishing, 2015. — P. 234–241. — https://doi.org/10.1007/978-3-319-24574-4_28

18. Shafiabadi M., Kamkar-Rouhani A., Ghavami Riabi S. R., et al. Identification of reservoir fractures on FMI image logs using Canny and Sobel edge detection algorithms // Oil & Gas Science and Technology – Revue d’IFP Energies nouvelles. — 2021. — Vol. 76. — P. 10. — https://doi.org/10.2516/ogst/2020086

19. Simard P. Y., Steinkraus D. and Platt J. C. Best practices for convolutional neural networks applied to visual document analysis // Seventh International Conference on Document Analysis and Recognition, 2003. Proceedings. Vol. 1. — IEEE Comput. Soc, 2003. — P. 958–963. — https://doi.org/10.1109/icdar.2003.1227801

20. Terven J., Córdova-Esparza D. M. and Romero-González J. A. A Comprehensive Review of YOLO Architectures in Computer Vision: From YOLOv1 to YOLOv8 and YOLO-NAS // Machine Learning and Knowledge Extraction. — 2023. — Vol. 5, no. 4. — P. 1680–1716. — https://doi.org/10.3390/make5040083

21. Vugrin K. W., Swiler L. P., Roberts R. M., et al. Confidence region estimation techniques for nonlinear regression in groundwater flow: Three case studies // Water Resources Research. — 2007. — Vol. 43, no. 3. — https://doi.org/10.1029/2005wr004804