Sea Ice Segmentation from SAR Data by Convolutional Transformer Networks

Sea ice is a crucial component of the Earth’s climate system and is highly sensitive to changes in temperature and atmospheric conditions. Accurate and timely measurement of sea ice parameters is important for understanding and predicting the impacts of climate change. Nevertheless, the amount of sa...

Full description

Bibliographic Details
Published in:IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium
Main Authors: Ristea, Nicolae-Cătălin, Anghel, Andrei, Datcu, Mihai
Format: Conference Object
Language:unknown
Published: 2023
Subjects:
EO Data Science
Sea ice
Online Access:https://elib.dlr.de/201616/
https://2023.ieeeigarss.org/
id ftdlr:oai:elib.dlr.de:201616
record_format openpolar
spelling ftdlr:oai:elib.dlr.de:201616 2024-05-19T07:48:13+00:00 Sea Ice Segmentation from SAR Data by Convolutional Transformer Networks Ristea, Nicolae-Cătălin Anghel, Andrei Datcu, Mihai 2023 https://elib.dlr.de/201616/ https://2023.ieeeigarss.org/ unknown Ristea, Nicolae-Cătălin und Anghel, Andrei und Datcu, Mihai (2023) Sea Ice Segmentation from SAR Data by Convolutional Transformer Networks. In: 2023 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2023, Seiten 168-171. IGARSS 2023, 2023-07-16 - 2023-07-21, Pasadena, CA, USA. doi:10.1109/IGARSS52108.2023.10283427 <https://doi.org/10.1109/IGARSS52108.2023.10283427>. ISBN 979-835032010-7. ISSN 2153-6996. EO Data Science Konferenzbeitrag PeerReviewed 2023 ftdlr https://doi.org/10.1109/IGARSS52108.2023.10283427 2024-04-25T01:11:02Z Sea ice is a crucial component of the Earth’s climate system and is highly sensitive to changes in temperature and atmospheric conditions. Accurate and timely measurement of sea ice parameters is important for understanding and predicting the impacts of climate change. Nevertheless, the amount of satellite data acquired over ice areas is huge, making the subjective measurements ineffective. Therefore, automated algorithms must be used in order to fully exploit the continuous data feeds coming from satellites. In this paper, we present a novel approach for sea ice segmentation based on SAR satellite imagery using hybrid convolutional transformer (ConvTr) networks. We show that our approach outperforms classical convolutional networks, while being considerably more efficient than pure transformer models. ConvTr obtained a mean intersection over union (mIoU) of 63.68% on the AI4Arctic data set, assuming an inference time of 120ms for a 400×400 km 2 product. Conference Object Sea ice German Aerospace Center: elib - DLR electronic library IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium 168 171
institution Open Polar
collection German Aerospace Center: elib - DLR electronic library
op_collection_id ftdlr
language unknown
topic EO Data Science
spellingShingle EO Data Science
Ristea, Nicolae-Cătălin
Anghel, Andrei
Datcu, Mihai
Sea Ice Segmentation from SAR Data by Convolutional Transformer Networks
topic_facet EO Data Science
description Sea ice is a crucial component of the Earth’s climate system and is highly sensitive to changes in temperature and atmospheric conditions. Accurate and timely measurement of sea ice parameters is important for understanding and predicting the impacts of climate change. Nevertheless, the amount of satellite data acquired over ice areas is huge, making the subjective measurements ineffective. Therefore, automated algorithms must be used in order to fully exploit the continuous data feeds coming from satellites. In this paper, we present a novel approach for sea ice segmentation based on SAR satellite imagery using hybrid convolutional transformer (ConvTr) networks. We show that our approach outperforms classical convolutional networks, while being considerably more efficient than pure transformer models. ConvTr obtained a mean intersection over union (mIoU) of 63.68% on the AI4Arctic data set, assuming an inference time of 120ms for a 400×400 km 2 product.
format Conference Object
author Ristea, Nicolae-Cătălin
Anghel, Andrei
Datcu, Mihai
author_facet Ristea, Nicolae-Cătălin
Anghel, Andrei
Datcu, Mihai
author_sort Ristea, Nicolae-Cătălin
title Sea Ice Segmentation from SAR Data by Convolutional Transformer Networks
title_short Sea Ice Segmentation from SAR Data by Convolutional Transformer Networks
title_full Sea Ice Segmentation from SAR Data by Convolutional Transformer Networks
title_fullStr Sea Ice Segmentation from SAR Data by Convolutional Transformer Networks
title_full_unstemmed Sea Ice Segmentation from SAR Data by Convolutional Transformer Networks
title_sort sea ice segmentation from sar data by convolutional transformer networks
publishDate 2023
url https://elib.dlr.de/201616/
https://2023.ieeeigarss.org/
genre Sea ice
genre_facet Sea ice
op_relation Ristea, Nicolae-Cătălin und Anghel, Andrei und Datcu, Mihai (2023) Sea Ice Segmentation from SAR Data by Convolutional Transformer Networks. In: 2023 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2023, Seiten 168-171. IGARSS 2023, 2023-07-16 - 2023-07-21, Pasadena, CA, USA. doi:10.1109/IGARSS52108.2023.10283427 <https://doi.org/10.1109/IGARSS52108.2023.10283427>. ISBN 979-835032010-7. ISSN 2153-6996.
op_doi https://doi.org/10.1109/IGARSS52108.2023.10283427
container_title IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium
container_start_page 168
op_container_end_page 171
_version_ 1799488752249208832

Similar Items