GeoCLR: georeference contrastive learning for efficient seafloor image interpretation
GeoCLR: georeference contrastive learning for efficient seafloor image interpretation
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 metres 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 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 ~86k images gathered using an Autonomous Underwater Vehicle (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.
1134 - 1155
Yamada, Takaki
81c66c35-0e2b-4342-80fa-cbee6ff9ce5f
Prugel-Bennett, Adam
b107a151-1751-4d8b-b8db-2c395ac4e14e
Pizarro, Oscar
45a78dfb-5f85-4595-bfc8-d1ac46f2e7bb
Williams, Stefan B.
c9477238-5139-4b74-804c-3b9b464f6949
Thornton, Blair
8293beb5-c083-47e3-b5f0-d9c3cee14be9
Yamada, Takaki
81c66c35-0e2b-4342-80fa-cbee6ff9ce5f
Prugel-Bennett, Adam
b107a151-1751-4d8b-b8db-2c395ac4e14e
Pizarro, Oscar
45a78dfb-5f85-4595-bfc8-d1ac46f2e7bb
Williams, Stefan B.
c9477238-5139-4b74-804c-3b9b464f6949
Thornton, Blair
8293beb5-c083-47e3-b5f0-d9c3cee14be9
Yamada, Takaki, Prugel-Bennett, Adam, Pizarro, Oscar, Williams, Stefan B. and Thornton, Blair
(2022)
GeoCLR: georeference contrastive learning for efficient seafloor image interpretation.
Field Robotics, 2, .
(doi:10.55417/fr.2022037).
Abstract
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 metres 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 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 ~86k images gathered using an Autonomous Underwater Vehicle (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.
Text
Yamada_2022_FR
- Accepted Manuscript
Text
Yamada_2022_FR
- Version of Record
More information
Accepted/In Press date: 20 April 2022
e-pub ahead of print date: 9 June 2022
Identifiers
Local EPrints ID: 456914
URI: http://eprints.soton.ac.uk/id/eprint/456914
PURE UUID: 394bbe96-f786-44fb-a66e-f7002dd60b2e
Catalogue record
Date deposited: 17 May 2022 16:37
Last modified: 17 Mar 2024 07:16
Export record
Altmetrics
Contributors
Author:
Takaki Yamada
Author:
Adam Prugel-Bennett
Author:
Oscar Pizarro
Author:
Stefan B. Williams
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics