A deep learning-based method to identify and count pelagic and mesopelagic fishes from trawl camera images
Fish counts and species information can be obtained from images taken within trawls, which enables trawl surveys to operate without extracting fish from their habitat, yields distribution data at fine scale for better interpretation of acoustic results, and can detect fish that are not retained in t...
Published in: | ICES Journal of Marine Science |
---|---|
Main Authors: | , , , |
Format: | Article in Journal/Newspaper |
Language: | English |
Published: |
Oxford University Press
2021
|
Subjects: | |
Online Access: | https://hdl.handle.net/11250/2833324 https://doi.org/10.1093/icesjms/fsab227 |
id |
ftunivbergen:oai:bora.uib.no:11250/2833324 |
---|---|
record_format |
openpolar |
spelling |
ftunivbergen:oai:bora.uib.no:11250/2833324 2023-05-15T17:47:06+02:00 A deep learning-based method to identify and count pelagic and mesopelagic fishes from trawl camera images Allken, Vaneeda Rosen, Shale Pettit Handegard, Nils Olav Malde, Ketil 2021 application/pdf https://hdl.handle.net/11250/2833324 https://doi.org/10.1093/icesjms/fsab227 eng eng Oxford University Press Norges forskningsråd: 270966 Norges forskningsråd: 309512 Norges forskningsråd: 203477 urn:issn:1054-3139 https://hdl.handle.net/11250/2833324 https://doi.org/10.1093/icesjms/fsab227 cristin:1961495 ICES Journal of Marine Science. 2021, fsab227. Navngivelse 4.0 Internasjonal http://creativecommons.org/licenses/by/4.0/deed.no Copyright International Council for the Exploration of the Sea 2021 fsab227 ICES Journal of Marine Science Journal article Peer reviewed 2021 ftunivbergen https://doi.org/10.1093/icesjms/fsab227 2023-03-14T17:42:44Z Fish counts and species information can be obtained from images taken within trawls, which enables trawl surveys to operate without extracting fish from their habitat, yields distribution data at fine scale for better interpretation of acoustic results, and can detect fish that are not retained in the catch due to mesh selection. To automate the process of image-based fish detection and identification, we trained a deep learning algorithm (RetinaNet) on images collected from the trawl-mounted Deep Vision camera system. In this study, we focused on the detection of blue whiting, Atlantic herring, Atlantic mackerel, and mesopelagic fishes from images collected in the Norwegian sea. To address the need for large amounts of annotated data to train these models, we used a combination of real and synthetic images, and obtained a mean average precision of 0.845 on a test set of 918 images. Regression models were used to compare predicted fish counts, which were derived from RetinaNet classification of fish in the individual image frames, with catch data collected at 20 trawl stations. We have automatically detected and counted fish from individual images, related these counts to the trawl catches, and discussed how to use this in regular trawl surveys. publishedVersion Article in Journal/Newspaper Norwegian Sea University of Bergen: Bergen Open Research Archive (BORA-UiB) Norwegian Sea ICES Journal of Marine Science 78 10 3780 3792 |
institution |
Open Polar |
collection |
University of Bergen: Bergen Open Research Archive (BORA-UiB) |
op_collection_id |
ftunivbergen |
language |
English |
description |
Fish counts and species information can be obtained from images taken within trawls, which enables trawl surveys to operate without extracting fish from their habitat, yields distribution data at fine scale for better interpretation of acoustic results, and can detect fish that are not retained in the catch due to mesh selection. To automate the process of image-based fish detection and identification, we trained a deep learning algorithm (RetinaNet) on images collected from the trawl-mounted Deep Vision camera system. In this study, we focused on the detection of blue whiting, Atlantic herring, Atlantic mackerel, and mesopelagic fishes from images collected in the Norwegian sea. To address the need for large amounts of annotated data to train these models, we used a combination of real and synthetic images, and obtained a mean average precision of 0.845 on a test set of 918 images. Regression models were used to compare predicted fish counts, which were derived from RetinaNet classification of fish in the individual image frames, with catch data collected at 20 trawl stations. We have automatically detected and counted fish from individual images, related these counts to the trawl catches, and discussed how to use this in regular trawl surveys. publishedVersion |
format |
Article in Journal/Newspaper |
author |
Allken, Vaneeda Rosen, Shale Pettit Handegard, Nils Olav Malde, Ketil |
spellingShingle |
Allken, Vaneeda Rosen, Shale Pettit Handegard, Nils Olav Malde, Ketil A deep learning-based method to identify and count pelagic and mesopelagic fishes from trawl camera images |
author_facet |
Allken, Vaneeda Rosen, Shale Pettit Handegard, Nils Olav Malde, Ketil |
author_sort |
Allken, Vaneeda |
title |
A deep learning-based method to identify and count pelagic and mesopelagic fishes from trawl camera images |
title_short |
A deep learning-based method to identify and count pelagic and mesopelagic fishes from trawl camera images |
title_full |
A deep learning-based method to identify and count pelagic and mesopelagic fishes from trawl camera images |
title_fullStr |
A deep learning-based method to identify and count pelagic and mesopelagic fishes from trawl camera images |
title_full_unstemmed |
A deep learning-based method to identify and count pelagic and mesopelagic fishes from trawl camera images |
title_sort |
deep learning-based method to identify and count pelagic and mesopelagic fishes from trawl camera images |
publisher |
Oxford University Press |
publishDate |
2021 |
url |
https://hdl.handle.net/11250/2833324 https://doi.org/10.1093/icesjms/fsab227 |
geographic |
Norwegian Sea |
geographic_facet |
Norwegian Sea |
genre |
Norwegian Sea |
genre_facet |
Norwegian Sea |
op_source |
fsab227 ICES Journal of Marine Science |
op_relation |
Norges forskningsråd: 270966 Norges forskningsråd: 309512 Norges forskningsråd: 203477 urn:issn:1054-3139 https://hdl.handle.net/11250/2833324 https://doi.org/10.1093/icesjms/fsab227 cristin:1961495 ICES Journal of Marine Science. 2021, fsab227. |
op_rights |
Navngivelse 4.0 Internasjonal http://creativecommons.org/licenses/by/4.0/deed.no Copyright International Council for the Exploration of the Sea 2021 |
op_doi |
https://doi.org/10.1093/icesjms/fsab227 |
container_title |
ICES Journal of Marine Science |
container_volume |
78 |
container_issue |
10 |
container_start_page |
3780 |
op_container_end_page |
3792 |
_version_ |
1766151406318780416 |