A real‐world dataset and data simulation algorithm for automated fish species identification

Abstract Developing high‐performing machine learning algorithms requires large amounts of annotated data. Manual annotation of data is labour‐intensive, and the cost and effort needed are an important obstacle to the development and deployment of automated analysis. In a previous work, we have shown...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Vaneeda Allken, Shale Rosen, Nils Olav Handegard, Ketil Malde
Formato:	article
Lenguaje:	EN
Publicado:	Wiley 2021
Materias:	data augmentation fish dataset machine learning synthetic data Meteorology. Climatology QC851-999 Geology QE1-996.5
Acceso en línea:	https://doaj.org/article/4ee5074aad3e440d80135a3951387236
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:4ee5074aad3e440d80135a3951387236
record_format	dspace
spelling	oai:doaj.org-article:4ee5074aad3e440d80135a39513872362021-11-23T13:44:38ZA real‐world dataset and data simulation algorithm for automated fish species identification2049-606010.1002/gdj3.114https://doaj.org/article/4ee5074aad3e440d80135a39513872362021-11-01T00:00:00Zhttps://doi.org/10.1002/gdj3.114https://doaj.org/toc/2049-6060Abstract Developing high‐performing machine learning algorithms requires large amounts of annotated data. Manual annotation of data is labour‐intensive, and the cost and effort needed are an important obstacle to the development and deployment of automated analysis. In a previous work, we have shown that deep learning classifiers can successfully be trained on synthetic images and annotations. Here, we provide a curated set of fish image data and backgrounds, the necessary software tools to generate synthetic images and annotations, and annotated real datasets to test classifier performance. The dataset is constructed from images collected using the Deep Vision system during two surveys from 2017 and 2018 that targeted economically important pelagic species in the Northeast Atlantic Ocean. We annotated a total of 1,879 images, randomly selected across trawl stations from both surveys, comprising 482 images of blue whiting, 456 images of Atlantic herring, 341 images of Atlantic mackerel, 335 images of mesopelagic fishes and 265 images containing a mixture of the four categories.Vaneeda AllkenShale RosenNils Olav HandegardKetil MaldeWileyarticledata augmentationfish datasetmachine learningsynthetic dataMeteorology. ClimatologyQC851-999GeologyQE1-996.5ENGeoscience Data Journal, Vol 8, Iss 2, Pp 199-209 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	data augmentation fish dataset machine learning synthetic data Meteorology. Climatology QC851-999 Geology QE1-996.5
spellingShingle	data augmentation fish dataset machine learning synthetic data Meteorology. Climatology QC851-999 Geology QE1-996.5 Vaneeda Allken Shale Rosen Nils Olav Handegard Ketil Malde A real‐world dataset and data simulation algorithm for automated fish species identification
description	Abstract Developing high‐performing machine learning algorithms requires large amounts of annotated data. Manual annotation of data is labour‐intensive, and the cost and effort needed are an important obstacle to the development and deployment of automated analysis. In a previous work, we have shown that deep learning classifiers can successfully be trained on synthetic images and annotations. Here, we provide a curated set of fish image data and backgrounds, the necessary software tools to generate synthetic images and annotations, and annotated real datasets to test classifier performance. The dataset is constructed from images collected using the Deep Vision system during two surveys from 2017 and 2018 that targeted economically important pelagic species in the Northeast Atlantic Ocean. We annotated a total of 1,879 images, randomly selected across trawl stations from both surveys, comprising 482 images of blue whiting, 456 images of Atlantic herring, 341 images of Atlantic mackerel, 335 images of mesopelagic fishes and 265 images containing a mixture of the four categories.
format	article
author	Vaneeda Allken Shale Rosen Nils Olav Handegard Ketil Malde
author_facet	Vaneeda Allken Shale Rosen Nils Olav Handegard Ketil Malde
author_sort	Vaneeda Allken
title	A real‐world dataset and data simulation algorithm for automated fish species identification
title_short	A real‐world dataset and data simulation algorithm for automated fish species identification
title_full	A real‐world dataset and data simulation algorithm for automated fish species identification
title_fullStr	A real‐world dataset and data simulation algorithm for automated fish species identification
title_full_unstemmed	A real‐world dataset and data simulation algorithm for automated fish species identification
title_sort	real‐world dataset and data simulation algorithm for automated fish species identification
publisher	Wiley
publishDate	2021
url	https://doaj.org/article/4ee5074aad3e440d80135a3951387236
work_keys_str_mv	AT vaneedaallken arealworlddatasetanddatasimulationalgorithmforautomatedfishspeciesidentification AT shalerosen arealworlddatasetanddatasimulationalgorithmforautomatedfishspeciesidentification AT nilsolavhandegard arealworlddatasetanddatasimulationalgorithmforautomatedfishspeciesidentification AT ketilmalde arealworlddatasetanddatasimulationalgorithmforautomatedfishspeciesidentification AT vaneedaallken realworlddatasetanddatasimulationalgorithmforautomatedfishspeciesidentification AT shalerosen realworlddatasetanddatasimulationalgorithmforautomatedfishspeciesidentification AT nilsolavhandegard realworlddatasetanddatasimulationalgorithmforautomatedfishspeciesidentification AT ketilmalde realworlddatasetanddatasimulationalgorithmforautomatedfishspeciesidentification
_version_	1718416700749119488

A real‐world dataset and data simulation algorithm for automated fish species identification

Ejemplares similares