Abstract
In TBM (Tunnel Boring Machine) construction process, the rock size analysis system plays an important role in assisting driving. Its core algorithm is based on semantic segmentation, and it brings challenges to dataset acquisition in real applications. To relieve this problem, this paper proposes a virtual-realistic fused dataset, short for ViRFD. The R-part is composed of a realistic dataset from our previous work, and the V-part is simulated by a learning-based method proposed in this paper. Unlike traditional manual methods, we use a virtual engine (Unity) to simulate datasets, since the corresponding ground-truth labels can be automatically extracted by the engine. Specifically, we propose a novel synthetic dataset simulator, named RockSegX. It contains abundant virtual 3D resources to ensure the diversity and fidelity of generated datasets. The main feature of RockSegX lies in its content flexibility, i.e., we are able to control the content of dataset by adjusting the values of several attributes. These attributes are carefully designed for reducing the content difference between V-part and R-part datasets. And we employ a learning-based method to automatically adjust the attributes so that the V-part dataset has the smallest content difference with the R-part. Experimental results show the effectiveness of our method in improving the quality of simulated dataset, and it further boosts the test accuracy for real-world segmentation.
Similar content being viewed by others
References
Anagnostou G, Kovári K (1996) Face stability conditions with earth-pressure-balanced shields. Tunn Undergr Space Technol 11(2):165–173
Barbosa IB, Cristani M, Caputo B, Rognhaugen A, Theoharis T (2018) Looking beyond appearances: synthetic training data for deep CNNS in re-identification. Comput Vis Image Underst 167:50–62
Brostow GJ, Fauqueur J, Cipolla R (2009) Semantic object classes in video: a high-definition ground truth database. Pattern Recogn Lett 30(2):88–97
Cordts M, Omran M, Ramos S, Rehfeld T, Enzweiler M, Benenson R, Franke U, Roth S, Schiele B (2016) The cityscapes dataset for semantic urban scene understanding. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 3213–3223
Cruz C, Foi A, Katkovnik V, Egiazarian K (2018) Nonlocality-reinforced convolutional neural networks for image denoising. IEEE Signal Process Lett 25(8):1216–1220. https://doi.org/10.1109/LSP.2018.2850222
Devaranjan J, Kar A, Fidler S (2020) Meta-sim2: unsupervised learning of scene structure for synthetic data generation. In: Proceeding of the European conference on computer vision, pp 715–733
Erben H (2016) Real-time material analysis and development of a collaboration and trading platform for mineral resources from underground construction projects. Doctoral thesis in Montanuniversity Leoben
Farrokh E, Rostami J (2008) Correlation of tunnel convergence with tbm operational parameters and chip size in the ghomroud tunnel, iran. Tunn Undergr Space Technol 23(6):700–710
Gaidon A, Wang Q, Cabon Y, Vig E (2016) Virtual worlds as proxy for multi-object tracking analysis. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 4340–4349
Geiger A, Lenz P, Stiller C, Urtasun R (2013) Vision meets robotics: the kitti dataset. Int J Robot Res 32(11):1231–1237
Geiger A, Lenz P, Urtasun R (2012) Are we ready for autonomous driving? The kitti vision benchmark suite. In: 2012 IEEE conference on computer vision and pattern recognition, pp 3354–3361. IEEE
Gong Q, Zhou X, Liu Y, Han B, Yin L (2021) Development of a real-time muck analysis system for assistant intelligence tbm tunnelling. Tunn Undergr Space Technol 107:103655
Gretton A, Borgwardt KM, Rasch MJ, Schölkopf B, Smola A (2012) A kernel two-sample test. J Mach Learn Res 13(1):723–773
Guyot O, Monredon T, Larosa D, Broussaud A (2004) Visiorock, an integrated vision technology for advanced control of aggregate circuits. Miner Eng 17(11–12):1227–1235
Hattori H, Naresh Boddeti V, Kitani K.M, Kanade T (2015) Learning scene-specific pedestrian detectors without real data. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 3819–3827
He K, Zhang X, Ren S, Sun J (2016) Deep residual learning for image recognition. In: 2016 IEEE conference on computer vision and pattern recognition (CVPR), pp 770–778
Heusel M, Ramsauer H, Unterthiner T, Nessler B, Hochreiter S (2017) Gans trained by a two time-scale update rule converge to a local nash equilibrium. Adv Neural Inf Process Syst 30
Hongxin W, Deming F (2007) Theoretical and test studies on balance control of epb shields. Chin Civil Eng J 40(5):61–68
Jiang F, Grigorev A, Rho S, Tian Z, Fu Y, Jifara W, Adil K, Liu S (2018) Medical image semantic segmentation based on deep learning. Neural Comput Appl 29:1257–1265
Juliani A, Berges V, Vckay E, Gao Y, Henry H, Mattar M, Lange D (2018) Unity: a general platform for intelligent agents. CoRR arxiv:1809.02627
Kang B, Lee Y, Nguyen TQ (2018) Depth-adaptive deep neural network for semantic segmentation. IEEE Trans Multimedia 20(9):2478–2490. https://doi.org/10.1109/TMM.2018.2798282
Kar A, Prakash A, Liu M.Y, Cameracci E, Yuan J, Rusiniak M, Acuna D, Torralba A, Fidler S (2019) Meta-sim: learning to generate synthetic datasets. In: 2019 IEEE/CVF International conference on computer vision (ICCV), pp 4550–4559
Kipf T.N, Welling M (2016) Semi-supervised classification with graph convolutional networks. arXiv preprint arXiv:1609.02907
Long J, Shelhamer E, Darrell T (2015) Fully convolutional networks for semantic segmentation. IEEE Trans Pattern Anal Mach Intell 39(4):640–651
Nie WZ, Jia WW, Li WH, Liu AA, Zhao SC (2021) 3D pose estimation based on reinforce learning for 2D image-based 3D model retrieval. IEEE Trans Multimed 23:1021–1034. https://doi.org/10.1109/TMM.2020.2991532
Nurzynska K, Iwaszenko S (2020) Application of texture features and machine learning methods to grain segmentation in rock material images. Image Anal Stereol 39(2):73–90
Outal S, Beucher S (2009) Controlling the ultimate opening residues for a robust delineation of fragmetned rocks. In: The 10th European Congress of Stereology and Image Analysis, Milan
Outal S, Jeulin D, Schleifer J (2011) A new method for estimating the 3d size-distribution curve of fragmented rocks out of 2d images. Image Anal Stereol 27(2):97–105
Peng C, Zhang X, Yu G, Luo G, Sun J (2017) Large kernel matters–improve semantic segmentation by global convolutional network. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 4353–4361
Pepik B, Stark M, Gehler P, Schiele B (2012) Teaching 3d geometry to deformable part models. In: 2012 IEEE conference on computer vision and pattern recognition, pp 3362–3369. IEEE
Richter S.R, Vineet V, Roth S, Koltun V (2016) Playing for data: ground truth from computer games. In: European conference on computer vision
Rispoli A, Ferrero AM, Cardu M, Farinetti A (2017) Determining the particle size of debris from a tunnel boring machine through photographic analysis and comparison between excavation performance and rock mass properties. Rock Mech Rock Eng 50(10):2805–2816
Ruiz N, Schulter S, Chandraker M (2019) Learning to simulate. In: International conference on learning representations
Satkin S, Lin J, Hebert M (2012) Data-driven scene understanding from 3d models. In: British machine vision conference, pp 128.1–128.11
Senniappan V, Subramanian J, Papageorgiou EI, Mohan S (2017) Application of fuzzy cognitive maps for crack categorization in columns of reinforced concrete structures. Neural Comput Appl 28:107–117
Shao C, Liao J, Li X, Su H (2015) An adaptive robust control for hard rock tunnel boring machine cutterhead driving system. In: ASME 2015 Dynamic systems and control conference, pp. V003T48A001–V003T48A001. American Society of Mechanical Engineers
Simonyan K, Zisserman A (2015) Very deep convolutional networks for large-scale image recognition. In: Proceedings of international conference on learning representations, pp 1–14
Smith B (2002) Improvements in blast fragmentation using measurement while drilling parameters. Fragblast 6(3/4):301–310
Sun X, Zheng L (2019) Dissecting person re-identification from the viewpoint of viewpoint. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp 608–617
Tremblay J, Prakash A, Acuna D, Brophy M, Jampani V, Anil C, To T, Cameracci E, Boochoon S, Birchfield S (2018) Training deep networks with synthetic data: Bridging the reality gap by domain randomization. In: Proceedings of the IEEE conference on computer vision and pattern recognition workshops, pp 969–977
Williams RJ (1992) Simple statistical gradient-following algorithms for connectionist reinforcement learning. Mach Learn 8(3–4):229–256
Xue Z, Chen L, Liu Z, Lin F, Mao W (2021) Rock segmentation visual system for assisting driving in tbm construction. Mach Vis Appl 32(4):1–12
Xue Z, Jia L, Sun W, Lin F, Liu Z, Mao W (2019) Multi mask learning of stone segmentation for auto-monitoring system in tbm construction. In: 2019 Chinese Control Conference (CCC), pp 8733–8738. 10.23919/ChiCC.2019.8865323
Xue Z, Mao W, Jiang W (2020) Ehanet: Efficient hybrid attention network towards real-time semantic segmentation. In: 2020 IEEE 6th International Conference on Computer and Communications (ICCC), pp 787–791. 10.1109/ICCC51575.2020.9345050
Xue Z, Mao W, Zheng L (2021) Learning to simulate complex scenes for street scene segmentation. IEEE Transactions on Multimedia p 1. 10.1109/TMM.2021.3062497
Yagiz S (2008) Utilizing rock mass properties for predicting tbm performance in hard rock condition. Tunn Undergr Space Technol 23(3):326–339
Yang H, Shi H, Gong G, Hu G (2009) Earth pressure balance control for epb shield. Sci China Ser E: Technol Sci 52(10):2840–2848
Yao Y, Zheng L, Yang X, Naphade M, Gedeon T (2020) Simulating content consistent vehicle datasets with attribute descent. In: Proceedings of European conference on computer vision, pp 775–791
Acknowledgements
This work was supported in part by the National Natural Science Foundation of China under Grant No. 61633019.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
We declare that we do not have any commercial or associative interest that represents a conflict of interest in connection with the work submitted.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix
Appendix
A. Details of attribute definition In RockSegX, we design 12 attributes to efficiently change the content of simulated dataset. These attributes all have great influence on the dataset appearance. Below shows the details of these attributes.
-
Number of small rocks, that changes the numbers of small rocks placed on the conveyor belt.
-
Mean of small rock scales, that changes the mean of small rock scales on the conveyor belt.
-
STD of small rock scales, that changes the standard deviation of small rock scales on the conveyor belt.
-
Number of large rocks, that changes the numbers of large rocks placed on the conveyor belt.
-
Mean of large rock scales, that changes the mean of large rock scales on the conveyor belt.
-
STD of large rock scales, that changes the standard deviation of large rock scales on the conveyor belt.
-
Critical value between bright and dark pixels, that determines the threshold of pixel value to be annotated as rock.
-
Light intensity, that changes the brightness of virtual environments.
-
Light rotation angle along X axis, that changes the light direction along X axis.
-
Light rotation angle along Y axis, that changes the light direction along Y axis.
-
Camera rotation angle along X axis, that changes the camera direction along X axis.
-
Camera rotation angle along Y axis, that changes the Camera direction along Y axis.
It is worth noting that the large rocks are those to be annotated as rock part in the images, and the small rocks are those to be ignored and annotated as background. The critical value between bright and dark pixels is of vital significance in dealing with invisible part in the images.
B. More visualization of simulated datasets The biggest feature of RockSegX lies in its flexibility in controlling the content of simulated datasets. Here, we give more visualization to show the simulated datasets by influence of changing attribute values. The figures are shown in Fig. 10.
Rights and permissions
About this article
Cite this article
Xue, Z., Chen, L., Liu, Z. et al. ViRFD: a virtual-realistic fused dataset for rock size analysis in TBM construction. Neural Comput & Applic 34, 13485–13498 (2022). https://doi.org/10.1007/s00521-022-07179-4
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00521-022-07179-4