An AUV Based Method for Estimating Hectare-scale Distributions of Deep Sea Cobalt-rich Manganese Crust Deposits

Neettiyath, Umesh; Thornton, Blair; Sangekar, Mehul; Nishida, Yuya; Ishii, Kazuo; Sato, Terukazu; Bodenmann, Adrian; Ura, Tamaki

doi:10.1109/oceanse.2019.8867481

Cited by 2 publications

(2 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, most algorithmic interpretations first convert images to lower-dimensional representations, or feature spaces, that can be more efficiently analyzed. Several types of feature descriptor have been investigated for seafloor image representations (Beijbom et al, 2012;Bewley et al, 2015;Kaeli and Singh, 2015;Neettiyath et al, 2021;Rao et al, 2017;Steinberg et al, 2011).…”

Section: Representation Learning For Seafloor Imagerymentioning

confidence: 99%

“…In Beijbom et al (2012) and Neettiyath et al (2021), color descriptors are designed to target known targets of scientific interest, such as corals (Beijbom et al, 2012) and mineral deposits (Neettiyath et al, 2021). Generic feature descriptors such as local binary patterns (LBPs) (Ojala et al, 2002) and sparse coding spatial pyramid matching (ScSPM) (Yang et al, 2009) have also been applied to capture multi-scale spatially invariant patterns in seafloor images (Bewley et al, 2015;Rao et al, 2017).…”

Section: Representation Learning For Seafloor Imagerymentioning

confidence: 99%

See 1 more Smart Citation

GeoCLR: Georeference Contrastive Learning for Efficient Seafloor Image Interpretation

Yamada¹,

Prügel-Bennett²,

Williams³

et al. 2022

Self Cite

View full text Add to dashboard Cite

This paper describes georeference contrastive learning of visual representation (GeoCLR) for efficient training of deep-learning convolutional neural networks (CNNs). The method leverages georeference information by generating a similar image pair using images taken of nearby locations, and contrasting these with an image pair that is far apart. The underlying assumption is that images gathered within a close distance are more likely to have similar visual appearance, where this can be reasonably satisfied in seafloor robotic imaging applications where image footprints are limited to edge lengths of a few meters and are taken so that they overlap along a vehicle’s trajectory, whereas seafloor substrates and habitats have patch sizes that are far larger. A key advantage of this method is that it is self-supervised and does not require any human input for CNN training. The method is computationally efficient, where results can be generated between dives during multi-day autonomous underwater vehicle (AUV) missions using computational resources that would be accessible during most oceanic field trials. We apply GeoCLR to habitat classification on a dataset that consists of ~86,000 images gathered using an AUV. We demonstrate how the latent representations generated by GeoCLR can be used to efficiently guide human annotation efforts, where the semi-supervised framework improves classification accuracy by an average of 10.2% compared to the state-of-the-art SimCLR using the same CNN and equivalent number of human annotations for training.

show abstract