Detecting, classifying, and counting blue whale calls with Siamese neural networks
International audience The goal of this project is to use acoustic signatures to detect, classify, and count the calls of four acoustic populations of blue whales so that, ultimately, the conservation status of each population can be better assessed. We used manual annotations from 350 h of audio re...
Published in: | The Journal of the Acoustical Society of America |
---|---|
Main Authors: | , , , , , , |
Other Authors: | , , , |
Format: | Article in Journal/Newspaper |
Language: | English |
Published: |
HAL CCSD
2021
|
Subjects: | |
Online Access: | https://hal.univ-brest.fr/hal-03263839 https://doi.org/10.1121/10.0004828 |
id |
ftanrparis:oai:HAL:hal-03263839v1 |
---|---|
record_format |
openpolar |
spelling |
ftanrparis:oai:HAL:hal-03263839v1 2024-09-15T18:00:02+00:00 Detecting, classifying, and counting blue whale calls with Siamese neural networks Zhong, Ming Torterotot, Maëlle Branch, Trevor Stafford, Kathleen Royer, Jean-Yves Dodhia, Rahul Lavista Ferres, Juan Laboratoire Géosciences Océan (LGO) Université de Bretagne Sud (UBS)-Institut Français de Recherche pour l'Exploitation de la Mer (IFREMER)-Université de Brest (UBO)-Centre National de la Recherche Scientifique (CNRS) Interdisciplinary Graduate School for the Blue plane ANR-17-EURE-0015,ISBlue,Interdisciplinary Graduate School for the Blue planet(2017) 2021-05 https://hal.univ-brest.fr/hal-03263839 https://doi.org/10.1121/10.0004828 en eng HAL CCSD Acoustical Society of America info:eu-repo/semantics/altIdentifier/doi/10.1121/10.0004828 hal-03263839 https://hal.univ-brest.fr/hal-03263839 doi:10.1121/10.0004828 ISSN: 0001-4966 EISSN: 1520-8524 Journal of the Acoustical Society of America https://hal.univ-brest.fr/hal-03263839 Journal of the Acoustical Society of America, 2021, 149 (5), pp.3086-3094. ⟨10.1121/10.0004828⟩ [SDE]Environmental Sciences info:eu-repo/semantics/article Journal articles 2021 ftanrparis https://doi.org/10.1121/10.0004828 2024-07-12T11:07:32Z International audience The goal of this project is to use acoustic signatures to detect, classify, and count the calls of four acoustic populations of blue whales so that, ultimately, the conservation status of each population can be better assessed. We used manual annotations from 350 h of audio recordings from the underwater hydrophones in the Indian Ocean to build a deep learning model to detect, classify, and count the calls from four acoustic song types. The method we used was Siamese neural networks (SNN), a class of neural network architectures that are used to find the similarity of the inputs by comparing their feature vectors, finding that they outperformed the more widely used convolutional neural networks (CNN). Specifically, the SNN outperform a CNN with 2% accuracy improvement in population classification and 1.7%-6.4% accuracy improvement in call count estimation for each blue whale population. In addition, even though we treat the call count estimation problem as a classification task and encode the number of calls in each spectrogram as a categorical variable, SNN surprisingly learned the ordinal relationship among them. SNN are robust and are shown here to be an effective way to automatically mine large acoustic datasets for blue whale calls. Article in Journal/Newspaper Blue whale Portail HAL-ANR (Agence Nationale de la Recherche) The Journal of the Acoustical Society of America 149 5 3086 3094 |
institution |
Open Polar |
collection |
Portail HAL-ANR (Agence Nationale de la Recherche) |
op_collection_id |
ftanrparis |
language |
English |
topic |
[SDE]Environmental Sciences |
spellingShingle |
[SDE]Environmental Sciences Zhong, Ming Torterotot, Maëlle Branch, Trevor Stafford, Kathleen Royer, Jean-Yves Dodhia, Rahul Lavista Ferres, Juan Detecting, classifying, and counting blue whale calls with Siamese neural networks |
topic_facet |
[SDE]Environmental Sciences |
description |
International audience The goal of this project is to use acoustic signatures to detect, classify, and count the calls of four acoustic populations of blue whales so that, ultimately, the conservation status of each population can be better assessed. We used manual annotations from 350 h of audio recordings from the underwater hydrophones in the Indian Ocean to build a deep learning model to detect, classify, and count the calls from four acoustic song types. The method we used was Siamese neural networks (SNN), a class of neural network architectures that are used to find the similarity of the inputs by comparing their feature vectors, finding that they outperformed the more widely used convolutional neural networks (CNN). Specifically, the SNN outperform a CNN with 2% accuracy improvement in population classification and 1.7%-6.4% accuracy improvement in call count estimation for each blue whale population. In addition, even though we treat the call count estimation problem as a classification task and encode the number of calls in each spectrogram as a categorical variable, SNN surprisingly learned the ordinal relationship among them. SNN are robust and are shown here to be an effective way to automatically mine large acoustic datasets for blue whale calls. |
author2 |
Laboratoire Géosciences Océan (LGO) Université de Bretagne Sud (UBS)-Institut Français de Recherche pour l'Exploitation de la Mer (IFREMER)-Université de Brest (UBO)-Centre National de la Recherche Scientifique (CNRS) Interdisciplinary Graduate School for the Blue plane ANR-17-EURE-0015,ISBlue,Interdisciplinary Graduate School for the Blue planet(2017) |
format |
Article in Journal/Newspaper |
author |
Zhong, Ming Torterotot, Maëlle Branch, Trevor Stafford, Kathleen Royer, Jean-Yves Dodhia, Rahul Lavista Ferres, Juan |
author_facet |
Zhong, Ming Torterotot, Maëlle Branch, Trevor Stafford, Kathleen Royer, Jean-Yves Dodhia, Rahul Lavista Ferres, Juan |
author_sort |
Zhong, Ming |
title |
Detecting, classifying, and counting blue whale calls with Siamese neural networks |
title_short |
Detecting, classifying, and counting blue whale calls with Siamese neural networks |
title_full |
Detecting, classifying, and counting blue whale calls with Siamese neural networks |
title_fullStr |
Detecting, classifying, and counting blue whale calls with Siamese neural networks |
title_full_unstemmed |
Detecting, classifying, and counting blue whale calls with Siamese neural networks |
title_sort |
detecting, classifying, and counting blue whale calls with siamese neural networks |
publisher |
HAL CCSD |
publishDate |
2021 |
url |
https://hal.univ-brest.fr/hal-03263839 https://doi.org/10.1121/10.0004828 |
genre |
Blue whale |
genre_facet |
Blue whale |
op_source |
ISSN: 0001-4966 EISSN: 1520-8524 Journal of the Acoustical Society of America https://hal.univ-brest.fr/hal-03263839 Journal of the Acoustical Society of America, 2021, 149 (5), pp.3086-3094. ⟨10.1121/10.0004828⟩ |
op_relation |
info:eu-repo/semantics/altIdentifier/doi/10.1121/10.0004828 hal-03263839 https://hal.univ-brest.fr/hal-03263839 doi:10.1121/10.0004828 |
op_doi |
https://doi.org/10.1121/10.0004828 |
container_title |
The Journal of the Acoustical Society of America |
container_volume |
149 |
container_issue |
5 |
container_start_page |
3086 |
op_container_end_page |
3094 |
_version_ |
1810437145030033408 |