Seafloor Pipeline Detection With Deep Learning
This paper presents fast, accurate, and automatic methods for detecting seafloor pipelines in multibeam echo sounder data with deep learning. The proposed methods take inspiration from the highly successful ResNet and YOLO deep learning models and tailor them to the idiosyncrasies of the seafloor pipeline detection task.
We use the area between lines and Hausdorff line distance functions to accurately evaluate how well methods can localize (pipe)lines. The same functions also show promise as loss functions compared to standard mean squared error, which does not include the regression variables' geometrical interpretation.
The model outperforms the highest likelihood baseline by more than 35% on a region-wise F1-score classification evaluation while being more than eight times more accurate than the baseline in locating pipelines. It is efficient, operating at over eighteen 32-ping image segments per second, which is far beyond real-time requirements.
P. E. Hagen, E. Borhaug, and Ø. Midtgaard, "Pipeline Inspection With Interferometric SAS", Sea Technology, vol. 51, no. 6, 2010.
J. Evans, P. Patron, B. Privat, et al., "AUTOTRACKER: Autonomous inspection - Capabilities and lessons learned in offshore operations", MTS/IEEE OCEANS, Biloxi, 2009. https://doi.org/10.23919/OCEANS.2009.5422339
Ø. Midtgaard, T. Krogstad, and P. E. Hagen, "Sonar detection and tracking of seafloor pipelines", in Proc UAM conf, 2011.
B. D. Van Veen. and K. M. Buckley, "Beamforming: a versatile approach to spatial filtering", IEEE ASSP Magazine, vol. 5, no. 2, pp. 4-24, 1988. https://doi.org/10.1109/53.665
K. He, X. Zhang, S. Ren, et al., "Deep residual learning for image recognition", Proc. CVPR, pp. 770-778, 2016. https://doi.org/10.1109/CVPR.2016.90
L. Bertinetto, J. Valmadre, J. F. Henriques, et al., "Fully-convolutional siamese networks for object tracking", LNCS, vol. 9914, pp. 850-865, 2016. https://arxiv.org/abs/1606.09549.
J. Redmon, S. Divvala, R. Girshick, et al., "You only look once: Uni ed, real-time object detection", Proc. CVPR, pp. 779-788, 2016. arXiv: 1506. 02640. https://doi.org/10.1109/CVPR.2016.91
W. Liu, D. Anguelov, D. Erhan, et al., "SSD: Single shot multibox detector", LNCS, vol. 9905, pp. 2137, 2016. arXiv: 1512.02325.
J. Redmon and A. Farhadi, "YOLO9000: Better, faster, stronger", Proc. CVPR, pp. 6517-6525, 2017. arXiv: 1612.08242. https://doi.org/10.1109/CVPR.2017.690
S. Wirtz and D. Paulus, "Evaluation of established line segment distance functions", 9th Open German-Russian Workshop on Pattern Recognition and Image Understanding, pp. 89-93, 2014.
R. Cipolla, Y. Gal, and A. Kendall, "Multitask Learning Using Uncertainty to Weigh Losses for Scene Geometry and Semantics", Proc. CVPR, pp. 7482-7491, 2018. doi: 10. 1109/CVPR.2018.00781. arXiv: 1705.07115.
O. Sener and V. Koltun, "Multi-task learning as multi-objective optimization", CoRR, vol. abs/1810.04650, 2018. arXiv: 1810.04650.
S. Ren, K. He, R. Girshick, et al., "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks", IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 39, no. 6, pp. 1137-1149, 2017. arXiv: 1506.01497. https://doi.org/10.1109/TPAMI.2016.2577031
V. N. Nguyen, R. Jenssen, and D. Roverso, "LS-Net : Fast Single-Shot Line-Segment Detector", arXiv: arXiv:1912.09532v2.
D. P. Kingma and J. L. Ba, "Adam: A method for stochastic optimization", in Proc. ICLR, 2015. arXiv: 1412.6980.
Copyright (c) 2021 Vemund Sigmundson Schøyen, Narada Dilp Warakagoda, Øivind Midtgaard
This work is licensed under a Creative Commons Attribution 4.0 International License.