One size fits all? Adaptation of trained CNNs to new marine acoustic environments
One size fits all? Adaptation of trained CNNs to new marine acoustic environments
Convolutional neural networks (CNNs) have the potential to enable a revolution in bioacoustics, allowing robust detection and classification of marine sound sources. As global Passive Acoustic Monitoring (PAM) datasets continue to expand it is critical we improve our confidence in the performance of models across different marine environments, if we are to exploit the full ecological value of information within the data. This work demonstrates the transferability of developed CNN models to new acoustic environments by using a pre-trained model developed for one location (West of Scotland, UK) and deploying it in a distinctly different soundscape (Gulf of Mexico, USA). In this work transfer learning is used to fine-tune an existing open-source ‘small-scale’ CNN, which detects odontocete tonal and broadband call types and vessel noise (operating between 0 and 48 kHz). The CNN is fine-tuned on training sets of differing sizes, from the unseen site, to understand the adaptability of a network to new marine acoustic environments. Fine-tuning with a small sample of site-specific data significantly improves the performance of the CNN in the new environment, across all classes. We demonstrate an improved performance in area-under-curve (AUC) score of 0.30, across four classes by fine-training with only 50 spectrograms per class, with a 5% improvement in accuracy between 50 frames and 500 frames. This work shows that only a small amount of site-specific data is needed to retrain a CNN, enabling researchers to harness the power of existing pre-trained models for their own datasets. The marine bioacoustic domain will benefit from a larger pool of global data for training large deep learning models, but we illustrate in this work that domain adaptation can be improved with limited site-specific exemplars.
Bioacoustics, Deep learning, Domain adaptation, Marine Mammal Detection, Marine acoustics, Soundscapes, Marine mammal detection
White, Ellen L.
1f019923-787e-4d89-9069-2f0b1ecf3506
Klinck, Holger
34ec2259-8bec-4264-bfcc-f5ea96126d2f
Bull, Jonathan M.
974037fd-544b-458f-98cc-ce8eca89e3c8
White, Paul R.
2dd2477b-5aa9-42e2-9d19-0806d994eaba
Risch, Denise
5b687ac5-831b-4640-bb88-5f03a27a3a27
December 2023
White, Ellen L.
1f019923-787e-4d89-9069-2f0b1ecf3506
Klinck, Holger
34ec2259-8bec-4264-bfcc-f5ea96126d2f
Bull, Jonathan M.
974037fd-544b-458f-98cc-ce8eca89e3c8
White, Paul R.
2dd2477b-5aa9-42e2-9d19-0806d994eaba
Risch, Denise
5b687ac5-831b-4640-bb88-5f03a27a3a27
White, Ellen L., Klinck, Holger, Bull, Jonathan M., White, Paul R. and Risch, Denise
(2023)
One size fits all? Adaptation of trained CNNs to new marine acoustic environments.
Ecological Informatics, 78, [102363].
(doi:10.1016/j.ecoinf.2023.102363).
Abstract
Convolutional neural networks (CNNs) have the potential to enable a revolution in bioacoustics, allowing robust detection and classification of marine sound sources. As global Passive Acoustic Monitoring (PAM) datasets continue to expand it is critical we improve our confidence in the performance of models across different marine environments, if we are to exploit the full ecological value of information within the data. This work demonstrates the transferability of developed CNN models to new acoustic environments by using a pre-trained model developed for one location (West of Scotland, UK) and deploying it in a distinctly different soundscape (Gulf of Mexico, USA). In this work transfer learning is used to fine-tune an existing open-source ‘small-scale’ CNN, which detects odontocete tonal and broadband call types and vessel noise (operating between 0 and 48 kHz). The CNN is fine-tuned on training sets of differing sizes, from the unseen site, to understand the adaptability of a network to new marine acoustic environments. Fine-tuning with a small sample of site-specific data significantly improves the performance of the CNN in the new environment, across all classes. We demonstrate an improved performance in area-under-curve (AUC) score of 0.30, across four classes by fine-training with only 50 spectrograms per class, with a 5% improvement in accuracy between 50 frames and 500 frames. This work shows that only a small amount of site-specific data is needed to retrain a CNN, enabling researchers to harness the power of existing pre-trained models for their own datasets. The marine bioacoustic domain will benefit from a larger pool of global data for training large deep learning models, but we illustrate in this work that domain adaptation can be improved with limited site-specific exemplars.
Text
OneSizeFitsAll_CNN_EcologicalInformatics
- Version of Record
More information
Accepted/In Press date: 1 November 2023
e-pub ahead of print date: 7 November 2023
Published date: December 2023
Additional Information:
Funding Information:
This work was supported by the Natural Environmental Research Council [grant number NE/S007210/1 ]. The COMPASS project has been supported by the EU's INTERREG VA Programme , managed by the Special EU Programmes Body . The views and opinions expressed in this document do not necessarily reflect those of the European Commission or the Special EU Programmes Body (SEUPB) .
Publisher Copyright:
© 2023 The Authors
Keywords:
Bioacoustics, Deep learning, Domain adaptation, Marine Mammal Detection, Marine acoustics, Soundscapes, Marine mammal detection
Identifiers
Local EPrints ID: 484457
URI: http://eprints.soton.ac.uk/id/eprint/484457
ISSN: 1574-9541
PURE UUID: 3b210aef-e55d-45f4-babf-962f82a1f066
Catalogue record
Date deposited: 16 Nov 2023 12:14
Last modified: 12 Jul 2024 01:34
Export record
Altmetrics
Contributors
Author:
Holger Klinck
Author:
Denise Risch
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
