North Atlantic right whale localization and recognition using very deep and leaky Neural Network
We describe a deep learning model that can be used to recognize individual right whales in aerial images. We developed our model using a data set provided by the National Oceanic and Atmospheric Administration. The main challenge we faced when working on this data set is that the size of the trainin...
Published in: | Mathematics for Application |
---|---|
Main Authors: | , |
Format: | Other/Unknown Material |
Language: | English |
Published: |
Vysoké učení technické v Brně, Fakulta strojního inženýrství, Ústav matematiky
2016
|
Subjects: | |
Online Access: | http://hdl.handle.net/11012/63787 https://doi.org/10.13164/ma.2016.11 |
id |
ftunivtbrno:oai:dspace.vutbr.cz:11012/63787 |
---|---|
record_format |
openpolar |
spelling |
ftunivtbrno:oai:dspace.vutbr.cz:11012/63787 2023-05-15T17:33:45+02:00 North Atlantic right whale localization and recognition using very deep and leaky Neural Network Kabani, A. El-Sakka, M. R. 2 5 2016 text 155-170 application/pdf http://hdl.handle.net/11012/63787 https://doi.org/10.13164/ma.2016.11 en eng Vysoké učení technické v Brně, Fakulta strojního inženýrství, Ústav matematiky Mathematics for Applications http://ma.fme.vutbr.cz/archiv/5_2/ma_5_2_kabani_el_sakka_final.pdf 1805-3629 doi:10.13164/ma.2016.11 http://hdl.handle.net/11012/63787 © Vysoké učení technické v Brně, Fakulta strojního inženýrství, Ústav matematiky openAccess Mathematics for Applications. 2016 vol. 5, č. 2, s. 155-170. ISSN 1805-3629 whale localization whale detection whale recognition deep learning convolutional neural network localization detection recognition image classi cation other publishedVersion 2016 ftunivtbrno https://doi.org/10.13164/ma.2016.11 2021-08-16T23:17:04Z We describe a deep learning model that can be used to recognize individual right whales in aerial images. We developed our model using a data set provided by the National Oceanic and Atmospheric Administration. The main challenge we faced when working on this data set is that the size of the training set is very small (4,544 images) with some classes having only 1 image. While this data set is by far the largest of its kind, it is very di cult to train a deep neural network with such a small data set. However, we were able to overcome this challenge by dividing this problem into smaller tasks and by reducing the viewpoint variance in the data set. First, we localize the body and the head of the whale using deep learning. Then, we align the whale and normalize it with respect to rotation. Finally, a network is used to recognize the whale by analyzing its callosities. The top-1 accuracy of the model is 69.7% and the top-5 accuracy is 85%. The solution we describe in this paper was ranked 5th (out of 364 teams) in a challenge to solve this problem. Other/Unknown Material North Atlantic North Atlantic right whale Brno University of Technology (VUT): Digital Library Mathematics for Application 5 2 155 170 |
institution |
Open Polar |
collection |
Brno University of Technology (VUT): Digital Library |
op_collection_id |
ftunivtbrno |
language |
English |
topic |
whale localization whale detection whale recognition deep learning convolutional neural network localization detection recognition image classi cation |
spellingShingle |
whale localization whale detection whale recognition deep learning convolutional neural network localization detection recognition image classi cation Kabani, A. El-Sakka, M. R. North Atlantic right whale localization and recognition using very deep and leaky Neural Network |
topic_facet |
whale localization whale detection whale recognition deep learning convolutional neural network localization detection recognition image classi cation |
description |
We describe a deep learning model that can be used to recognize individual right whales in aerial images. We developed our model using a data set provided by the National Oceanic and Atmospheric Administration. The main challenge we faced when working on this data set is that the size of the training set is very small (4,544 images) with some classes having only 1 image. While this data set is by far the largest of its kind, it is very di cult to train a deep neural network with such a small data set. However, we were able to overcome this challenge by dividing this problem into smaller tasks and by reducing the viewpoint variance in the data set. First, we localize the body and the head of the whale using deep learning. Then, we align the whale and normalize it with respect to rotation. Finally, a network is used to recognize the whale by analyzing its callosities. The top-1 accuracy of the model is 69.7% and the top-5 accuracy is 85%. The solution we describe in this paper was ranked 5th (out of 364 teams) in a challenge to solve this problem. |
format |
Other/Unknown Material |
author |
Kabani, A. El-Sakka, M. R. |
author_facet |
Kabani, A. El-Sakka, M. R. |
author_sort |
Kabani, A. |
title |
North Atlantic right whale localization and recognition using very deep and leaky Neural Network |
title_short |
North Atlantic right whale localization and recognition using very deep and leaky Neural Network |
title_full |
North Atlantic right whale localization and recognition using very deep and leaky Neural Network |
title_fullStr |
North Atlantic right whale localization and recognition using very deep and leaky Neural Network |
title_full_unstemmed |
North Atlantic right whale localization and recognition using very deep and leaky Neural Network |
title_sort |
north atlantic right whale localization and recognition using very deep and leaky neural network |
publisher |
Vysoké učení technické v Brně, Fakulta strojního inženýrství, Ústav matematiky |
publishDate |
2016 |
url |
http://hdl.handle.net/11012/63787 https://doi.org/10.13164/ma.2016.11 |
op_coverage |
2 5 |
genre |
North Atlantic North Atlantic right whale |
genre_facet |
North Atlantic North Atlantic right whale |
op_source |
Mathematics for Applications. 2016 vol. 5, č. 2, s. 155-170. ISSN 1805-3629 |
op_relation |
Mathematics for Applications http://ma.fme.vutbr.cz/archiv/5_2/ma_5_2_kabani_el_sakka_final.pdf 1805-3629 doi:10.13164/ma.2016.11 http://hdl.handle.net/11012/63787 |
op_rights |
© Vysoké učení technické v Brně, Fakulta strojního inženýrství, Ústav matematiky openAccess |
op_doi |
https://doi.org/10.13164/ma.2016.11 |
container_title |
Mathematics for Application |
container_volume |
5 |
container_issue |
2 |
container_start_page |
155 |
op_container_end_page |
170 |
_version_ |
1766132346385334272 |