Using Deep Learning to Count Albatrosses from Space

In this paper we test the use of a deep learning approach to automatically count Wandering Albatrosses in Very High Resolution (VHR) satellite imagery. We use a dataset of manually labelled imagery provided by the British Antarctic Survey to train and develop our methods. We employ a U-Net architect...

Full description

Bibliographic Details
Main Authors: Bowler, Ellen, Fretwell, Peter T., French, Geoffrey, Mackiewicz, Michal
Format: Article in Journal/Newspaper
Language:unknown
Published: arXiv 2019
Subjects:
Computer Vision and Pattern Recognition cs.CV
FOS Computer and information sciences
Antarctic
Antarc*
British Antarctic Survey
Online Access:https://dx.doi.org/10.48550/arxiv.1907.02040
https://arxiv.org/abs/1907.02040
id ftdatacite:10.48550/arxiv.1907.02040
record_format openpolar
spelling ftdatacite:10.48550/arxiv.1907.02040 2023-05-15T13:34:05+02:00 Using Deep Learning to Count Albatrosses from Space Bowler, Ellen Fretwell, Peter T. French, Geoffrey Mackiewicz, Michal 2019 https://dx.doi.org/10.48550/arxiv.1907.02040 https://arxiv.org/abs/1907.02040 unknown arXiv arXiv.org perpetual, non-exclusive license http://arxiv.org/licenses/nonexclusive-distrib/1.0/ Computer Vision and Pattern Recognition cs.CV FOS Computer and information sciences Article CreativeWork article Preprint 2019 ftdatacite https://doi.org/10.48550/arxiv.1907.02040 2022-03-10T16:41:39Z In this paper we test the use of a deep learning approach to automatically count Wandering Albatrosses in Very High Resolution (VHR) satellite imagery. We use a dataset of manually labelled imagery provided by the British Antarctic Survey to train and develop our methods. We employ a U-Net architecture, designed for image segmentation, to simultaneously classify and localise potential albatrosses. We aid training with the use of the Focal Loss criterion, to deal with extreme class imbalance in the dataset. Initial results achieve peak precision and recall values of approximately 80%. Finally we assess the model's performance in relation to inter-observer variation, by comparing errors against an image labelled by multiple observers. We conclude model accuracy falls within the range of human counters. We hope that the methods will streamline the analysis of VHR satellite images, enabling more frequent monitoring of a species which is of high conservation concern. : 4 pages, 5 figures, to be presented at IEEE 2019 International Geoscience & Remote Sensing Symposium (IGARSS 2019), scheduled for July 28 - August 2, 2019 Article in Journal/Newspaper Antarc* Antarctic British Antarctic Survey DataCite Metadata Store (German National Library of Science and Technology) Antarctic
institution Open Polar
collection DataCite Metadata Store (German National Library of Science and Technology)
op_collection_id ftdatacite
language unknown
topic Computer Vision and Pattern Recognition cs.CV
FOS Computer and information sciences
spellingShingle Computer Vision and Pattern Recognition cs.CV
FOS Computer and information sciences
Bowler, Ellen
Fretwell, Peter T.
French, Geoffrey
Mackiewicz, Michal
Using Deep Learning to Count Albatrosses from Space
topic_facet Computer Vision and Pattern Recognition cs.CV
FOS Computer and information sciences
description In this paper we test the use of a deep learning approach to automatically count Wandering Albatrosses in Very High Resolution (VHR) satellite imagery. We use a dataset of manually labelled imagery provided by the British Antarctic Survey to train and develop our methods. We employ a U-Net architecture, designed for image segmentation, to simultaneously classify and localise potential albatrosses. We aid training with the use of the Focal Loss criterion, to deal with extreme class imbalance in the dataset. Initial results achieve peak precision and recall values of approximately 80%. Finally we assess the model's performance in relation to inter-observer variation, by comparing errors against an image labelled by multiple observers. We conclude model accuracy falls within the range of human counters. We hope that the methods will streamline the analysis of VHR satellite images, enabling more frequent monitoring of a species which is of high conservation concern. : 4 pages, 5 figures, to be presented at IEEE 2019 International Geoscience & Remote Sensing Symposium (IGARSS 2019), scheduled for July 28 - August 2, 2019
format Article in Journal/Newspaper
author Bowler, Ellen
Fretwell, Peter T.
French, Geoffrey
Mackiewicz, Michal
author_facet Bowler, Ellen
Fretwell, Peter T.
French, Geoffrey
Mackiewicz, Michal
author_sort Bowler, Ellen
title Using Deep Learning to Count Albatrosses from Space
title_short Using Deep Learning to Count Albatrosses from Space
title_full Using Deep Learning to Count Albatrosses from Space
title_fullStr Using Deep Learning to Count Albatrosses from Space
title_full_unstemmed Using Deep Learning to Count Albatrosses from Space
title_sort using deep learning to count albatrosses from space
publisher arXiv
publishDate 2019
url https://dx.doi.org/10.48550/arxiv.1907.02040
https://arxiv.org/abs/1907.02040
geographic Antarctic
geographic_facet Antarctic
genre Antarc*
Antarctic
British Antarctic Survey
genre_facet Antarc*
Antarctic
British Antarctic Survey
op_rights arXiv.org perpetual, non-exclusive license
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
op_doi https://doi.org/10.48550/arxiv.1907.02040
_version_ 1766048768516423680

Similar Items