Comparison of Deep Learning Models for the Classification of Noctilucent Cloud Images
Optically thin layers of tiny ice particles near the summer mesopause, known as noctilucent clouds, are of significant interest within the aeronomy and climate science communities. Groundbased optical cameras mounted at various locations in the arctic regions collect the dataset during favorable sum...
Published in: | Remote Sensing |
---|---|
Main Authors: | , , |
Format: | Article in Journal/Newspaper |
Language: | English |
Published: |
MDPI
2022
|
Subjects: | |
Online Access: | https://hdl.handle.net/10037/27344 https://doi.org/10.3390/rs14102306 |
id |
ftunivtroemsoe:oai:munin.uit.no:10037/27344 |
---|---|
record_format |
openpolar |
spelling |
ftunivtroemsoe:oai:munin.uit.no:10037/27344 2023-05-15T15:06:15+02:00 Comparison of Deep Learning Models for the Classification of Noctilucent Cloud Images Sapkota, Rajendra Sharma, Puneet Mann, Ingrid 2022-05-10 https://hdl.handle.net/10037/27344 https://doi.org/10.3390/rs14102306 eng eng MDPI Remote Sensing Sapkota, Sharma, Mann. Comparison of Deep Learning Models for the Classification of Noctilucent Cloud Images. Remote Sensing. 2022;14(10) FRIDAID 2045685 doi:10.3390/rs14102306 2072-4292 https://hdl.handle.net/10037/27344 Attribution 4.0 International (CC BY 4.0) openAccess Copyright 2022 The Author(s) https://creativecommons.org/licenses/by/4.0 CC-BY Journal article Tidsskriftartikkel Peer reviewed publishedVersion 2022 ftunivtroemsoe https://doi.org/10.3390/rs14102306 2022-11-17T00:01:31Z Optically thin layers of tiny ice particles near the summer mesopause, known as noctilucent clouds, are of significant interest within the aeronomy and climate science communities. Groundbased optical cameras mounted at various locations in the arctic regions collect the dataset during favorable summer times. In this paper, first, we compare the performances of various deep learningbased image classifiers against a baseline machine learning model trained with support vector machine (SVM) algorithm to identify an effective and lightweight model for the classification of noctilucent clouds. The SVM classifier is trained with histogram of oriented gradient (HOG) features, and deep learning models such as SqueezeNet, ShuffleNet, MobileNet, and Resnet are fine-tuned based on the dataset. The dataset includes images observed from different locations in northern Europe with varied weather conditions. Second, we investigate the most informative pixels for the classification decision on test images. The pixel-level attributions calculated using the guide back-propagation algorithm are visualized as saliency maps. Our results indicate that the SqueezeNet model achieves an F1 score of 0.95. In addition, SqueezeNet is the lightest model used in our experiments, and the saliency maps obtained for a set of test images correspond better with relevant regions associated with noctilucent clouds. Article in Journal/Newspaper Arctic University of Tromsø: Munin Open Research Archive Arctic Remote Sensing 14 10 2306 |
institution |
Open Polar |
collection |
University of Tromsø: Munin Open Research Archive |
op_collection_id |
ftunivtroemsoe |
language |
English |
description |
Optically thin layers of tiny ice particles near the summer mesopause, known as noctilucent clouds, are of significant interest within the aeronomy and climate science communities. Groundbased optical cameras mounted at various locations in the arctic regions collect the dataset during favorable summer times. In this paper, first, we compare the performances of various deep learningbased image classifiers against a baseline machine learning model trained with support vector machine (SVM) algorithm to identify an effective and lightweight model for the classification of noctilucent clouds. The SVM classifier is trained with histogram of oriented gradient (HOG) features, and deep learning models such as SqueezeNet, ShuffleNet, MobileNet, and Resnet are fine-tuned based on the dataset. The dataset includes images observed from different locations in northern Europe with varied weather conditions. Second, we investigate the most informative pixels for the classification decision on test images. The pixel-level attributions calculated using the guide back-propagation algorithm are visualized as saliency maps. Our results indicate that the SqueezeNet model achieves an F1 score of 0.95. In addition, SqueezeNet is the lightest model used in our experiments, and the saliency maps obtained for a set of test images correspond better with relevant regions associated with noctilucent clouds. |
format |
Article in Journal/Newspaper |
author |
Sapkota, Rajendra Sharma, Puneet Mann, Ingrid |
spellingShingle |
Sapkota, Rajendra Sharma, Puneet Mann, Ingrid Comparison of Deep Learning Models for the Classification of Noctilucent Cloud Images |
author_facet |
Sapkota, Rajendra Sharma, Puneet Mann, Ingrid |
author_sort |
Sapkota, Rajendra |
title |
Comparison of Deep Learning Models for the Classification of Noctilucent Cloud Images |
title_short |
Comparison of Deep Learning Models for the Classification of Noctilucent Cloud Images |
title_full |
Comparison of Deep Learning Models for the Classification of Noctilucent Cloud Images |
title_fullStr |
Comparison of Deep Learning Models for the Classification of Noctilucent Cloud Images |
title_full_unstemmed |
Comparison of Deep Learning Models for the Classification of Noctilucent Cloud Images |
title_sort |
comparison of deep learning models for the classification of noctilucent cloud images |
publisher |
MDPI |
publishDate |
2022 |
url |
https://hdl.handle.net/10037/27344 https://doi.org/10.3390/rs14102306 |
geographic |
Arctic |
geographic_facet |
Arctic |
genre |
Arctic |
genre_facet |
Arctic |
op_relation |
Remote Sensing Sapkota, Sharma, Mann. Comparison of Deep Learning Models for the Classification of Noctilucent Cloud Images. Remote Sensing. 2022;14(10) FRIDAID 2045685 doi:10.3390/rs14102306 2072-4292 https://hdl.handle.net/10037/27344 |
op_rights |
Attribution 4.0 International (CC BY 4.0) openAccess Copyright 2022 The Author(s) https://creativecommons.org/licenses/by/4.0 |
op_rightsnorm |
CC-BY |
op_doi |
https://doi.org/10.3390/rs14102306 |
container_title |
Remote Sensing |
container_volume |
14 |
container_issue |
10 |
container_start_page |
2306 |
_version_ |
1766337901113638912 |