Semi-universal geo-crack detection by machine learning

被引:5
作者
Shi, Yongxiang [1 ,2 ]
Ballesio, Marco [2 ]
Johansen, Kasper [2 ]
Trentman, Daniel [2 ]
Huang, Yunsong [2 ]
McCabe, Matthew F. [2 ]
Bruhn, Ronald [3 ]
Schuster, Gerard [2 ,3 ]
机构
[1] Peking Univ, Sch Earth & Space Sci, Beijing, Peoples R China
[2] King Abdullah Univ Sci & Technol, Thuwal, Makkah, Saudi Arabia
[3] Univ Utah, Geol & Geophys Dept, Salt Lake City, UT 84112 USA
关键词
geo-crack detection; rock cracks; U-Net; convolutional neural network; transfer learning; machine learning; CONVOLUTIONAL NEURAL-NETWORKS; U-NET;
D O I
10.3389/feart.2023.1073211
中图分类号
P [天文学、地球科学];
学科分类号
07 ;
摘要
Introduction: Cracks are a key feature that determines the structural integrity of rocks, and their angular distribution can be used to determine the local or regional stress patterns. The temporal growth of cracks can be monitored in order to predict impending failures of materials or structures such as a weakened dam. Thus, cracks and their spatial-temporal distributions should be automatically monitored for assessing their structural integrity, the associated stress patterns and their potential for failure.Method: We show that the U-Net convolutional neural network, semantic segmentation and transfer learning can be used to accurately detect cracks in drone photos of sedimentary massifs. In this case, the crack distributions are used to assess the safest areas for tunnel excavation. Compared to the coarse performance of ridge detection, the U-Net accuracy in identifying cracks in images can be as high as 98% when evaluated against human identification, which is sufficient for assessing the general crack properties of the rock faces for the engineering project.Result: Based on approximately 100 h of manual cracks labeling in 127 drone photos and 20 h of network training, the U-Net was able to successfully detect cracks in 23,845 high-resolution photographs in less than 22 h using two Nvidia V100 GPUs. Meanwhile, the network was able to detect more than 80% of the observable cracks of a volcanic outcrop in Idaho without additional training. With a modest amount of extra labeling on photos of the volcanic outcrop and transfer training, we found that the accuracy significantly improved. The surprising outcome of this research is that the U-Net crack detector laboriously trained on photos of sedimentary rocks can also be effectively applied to photos of volcanic rock faces. This can be important for real-time assessment of geological hazards and lithology information for dam inspection and planetary exploration by autonomous vehicles. For another application, we accurately detected fractures and faults with a scale of tens of kilometers from Martian photographs.Conclusions: In summary, our methodology of using CNN with transfer training suggests that it can be used as a semi-universal detector of cracks in across a range of diverse geological settings.
引用
收藏
页数:13
相关论文
共 52 条
[1]   Learning from Imbalanced Data Sets with Weighted Cross-Entropy Function [J].
Aurelio, Yuri Sousa ;
de Almeida, Gustavo Matheus ;
de Castro, Cristiano Leite ;
Braga, Antonio Padua .
NEURAL PROCESSING LETTERS, 2019, 50 (02) :1937-1949
[2]   A New Rock Mass Quality Rating System: Rock Mass Quality Rating (RMQR) and Its Application to the Estimation of Geomechanical Characteristics of Rock Masses [J].
Aydan, Oemer ;
Ulusay, Resat ;
Tokashiki, Naohiko .
ROCK MECHANICS AND ROCK ENGINEERING, 2014, 47 (04) :1255-1276
[3]  
Behboodi B, 2019, IEEE ENG MED BIO, P6628, DOI 10.1109/EMBC.2019.8857218
[4]   Ground-based and UAV-Based photogrammetry: A multi-scale, high-resolution mapping tool for structural geology and paleoseismology [J].
Bemis, Sean P. ;
Micklethwaite, Steven ;
Turner, Darren ;
James, Mike R. ;
Akciz, Sinan ;
Thiele, Sam T. ;
Bangash, Hasnain Ali .
JOURNAL OF STRUCTURAL GEOLOGY, 2014, 69 :163-178
[5]   Autonomous concrete crack detection using deep fully convolutional neural network [J].
Cao Vu Dung ;
Le Duc Anh .
AUTOMATION IN CONSTRUCTION, 2019, 99 :52-58
[6]   Deep Learning-Based Crack Damage Detection Using Convolutional Neural Networks [J].
Cha, Young-Jin ;
Choi, Wooram ;
Buyukozturk, Oral .
COMPUTER-AIDED CIVIL AND INFRASTRUCTURE ENGINEERING, 2017, 32 (05) :361-378
[7]  
Cheng JR, 2018, TENCON IEEE REGION, P0462, DOI 10.1109/TENCON.2018.8650059
[8]  
Constantin A, 2018, 2018 IEEE ASIA PACIFIC CONFERENCE ON CIRCUITS AND SYSTEMS (APCCAS 2018), P423, DOI 10.1109/APCCAS.2018.8605652
[9]  
Daniel M., 2003, J GEOPHYS RES-PLANET, V108
[10]   Comparison of deep convolutional neural networks and edge detectors for image-based crack detection in concrete [J].
Dorafshan, Sattar ;
Thomas, Robert J. ;
Maguire, Marc .
CONSTRUCTION AND BUILDING MATERIALS, 2018, 186 :1031-1045