Semantic Segmentation of Underwater Sonar Imagery with Deep Learning

Rahnemoonfar, Maryam; Dobbs, Dugan

doi:10.1109/igarss.2019.8898742

Cited by 26 publications

(12 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Note especially that in the lowest training data scenario of five images, IDSS provides the largest improvement (0.128 MPA) over the second best method. Note also that Rahnemoonfar, et al [29] does not perform particularly well on our SAS dataset. We conjecture this for a couple of reasons: 1.)…”

Section: E Computational Burdenmentioning

confidence: 68%

“…There has been recent work in deep-networks [29], [30] trained for sidescan sonar images to segment seabed environment textures. These models are trained in a purely supervised fashion (using tedious human annotation), with training and evaluation over a relatively homogeneous image collection.…”

Section: Related Workmentioning

confidence: 99%

“…their approach was designed for sidescan sonar and 2.) being a purely supervised method, the method in [29] is highly reliant on generous training, which is practically infeasible and not available in our SAS dataset.…”

Section: E Computational Burdenmentioning

confidence: 99%

“…The IDSS network and the supervised U-Net are architecturally identical but of course with distinct optimized parameters. We also design Rahnemoonfar et al[29] consistent with the description in their paper and to have roughly the same number of parameters as the supervised U-Net.…”

mentioning

confidence: 99%

See 3 more Smart Citations

Iterative, Deep Synthetic Aperture Sonar Image Segmentation

Sun,

Gerg,

Monga

2022

Preprint

View full text Add to dashboard Cite

Synthetic aperture sonar (SAS) systems produce high-resolution images of the seabed environment. Moreover, deep learning has demonstrated superior ability in finding robust features for automating imagery analysis. However, the success of deep learning is conditioned on having lots of labeled training data, but obtaining generous pixel-level annotations of SAS imagery is often practically infeasible. This challenge has thus far limited the adoption of deep learning methods for SAS segmentation. Algorithms exist to segment SAS imagery in an unsupervised manner, but they lack the benefit of state-of-theart learning methods and the results present significant room for improvement. In view of the above, we propose a new iterative algorithm for unsupervised SAS image segmentation combining superpixel formation, deep learning, and traditional clustering methods. We call our method Iterative Deep Unsupervised Segmentation (IDUS). IDUS is an unsupervised learning framework that can be divided into four main steps: 1) A deep network estimates class assignments. 2) Low-level image features from the deep network are clustered into superpixels. 3) Superpixels are clustered into class assignments (which we call pseudo-labels) using k-means. 4) Resulting pseudo-labels are used for loss backpropagation of the deep network prediction. These four steps are performed iteratively until convergence. A comparison of IDUS to current state-of-the-art methods on a realistic benchmark dataset for SAS image segmentation demonstrates the benefits of our proposal even as the IDUS incurs a much lower computational burden during inference (actual labeling of a test image). Because our design combines merits of classical superpixel methods with deep learning, practically we demonstrate a very significant benefit in terms of reduced selection bias, i.e. IDUS shows markedly improved robustness against the choice of training images. Finally, we also develop a semi-supervised (SS) extension of IDUS called IDSS and demonstrate experimentally that it can further enhance performance while outperforming supervised alternatives that exploit the same labeled training imagery.

show abstract

Section: E Computational Burdenmentioning

confidence: 68%

Section: Related Workmentioning

confidence: 99%

Section: E Computational Burdenmentioning

confidence: 99%

mentioning

confidence: 99%

See 2 more Smart Citations

Iterative, Deep Synthetic Aperture Sonar Image Segmentation

Sun,

Gerg,

Monga

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…The advancement of AI techniques in recent years has had a great impact on our approaches to data analysis. Deep learning, in particular, has shown great success in many areas of practical interest such as classification (Krizhevsky and others, 2012a;Szegedy and others, 2015a;Sheppard and Rahnemoonfar, 2017), object recognition (Girshick and others, 2014;Hariharan and others, 2014), counting Sheppard, 2017a, 2017;Rahnemoonfar and others, 2019a) and semantic segmentation (Farabet and others, 2013;Mostajabi and others, 2015;Rahnemoonfar and others, 2018;Rahnemoonfar and Dobbs, 2019). Despite their progress, these algorithms are limited mainly to optical imagery.…”

Section: Introductionmentioning

confidence: 99%

Deep multi-scale learning for automatic tracking of internal layers of ice in radar data

et al. 2020

Self Cite

View full text Add to dashboard Cite

In this study, our goal is to track internal ice layers on the Snow Radar data collected by NASA Operation IceBridge. We examine the application of deep learning methods on radar data gathered from polar regions. Artificial intelligence techniques have displayed impressive success in many practical fields. Deep neural networks owe their success to the availability of massive labeled data. However, in many real-world problems, even when a large dataset is available, deep learning methods have shown less success, due to causes such as lack of a large labeled dataset, presence of noise in the data or missing data. In our radar data, the presence of noise is one of the main obstacles in utilizing popular deep learning methods such as transfer learning. Our experiments show that if the neural network is trained to detect contours of objects in electro-optical imagery, it can only track a low percentage of contours in radar data. Fine-tuning and further training do not provide any better results. However, we show that selecting the right model and training it on the radar imagery from the start yields far better results.

show abstract

Environmental Target Recognition System for ROV Based on Forward Looking Sonar

Liu

et al. 2023

Lecture Notes in Electrical Engineering

View full text Add to dashboard Cite

Semantic Segmentation of Underwater Sonar Imagery with Deep Learning

Cited by 26 publications

References 11 publications

Iterative, Deep Synthetic Aperture Sonar Image Segmentation

Iterative, Deep Synthetic Aperture Sonar Image Segmentation

Deep multi-scale learning for automatic tracking of internal layers of ice in radar data

Environmental Target Recognition System for ROV Based on Forward Looking Sonar

Contact Info

Product

Resources

About