A weakly supervised transfer learning approach for radar sounder data segmentation
Airborne radar sounders (RSs) are active sensors that acquire subsurface data for Earth observation. RS data (radargrams) provide information on buried geology by imaging subsurface dielectric discontinuities. Recently, several automatic RS target identification techniques have been proposed, with c...
Published in: | IEEE Transactions on Geoscience and Remote Sensing |
---|---|
Main Authors: | , , |
Other Authors: | , , |
Format: | Article in Journal/Newspaper |
Language: | English |
Published: |
2023
|
Subjects: | |
Online Access: | https://hdl.handle.net/11582/336789 https://doi.org/10.1109/TGRS.2023.3252939 https://ieeexplore.ieee.org/document/10059004 |
id |
ftfbkssleriris:oai:cris.fbk.eu:11582/336789 |
---|---|
record_format |
openpolar |
spelling |
ftfbkssleriris:oai:cris.fbk.eu:11582/336789 2024-09-15T17:44:31+00:00 A weakly supervised transfer learning approach for radar sounder data segmentation Miguel Hoyo Garcia Elena Donini Francesca Bovolo Hoyo Garcia, Miguel Donini, Elena Bovolo, Francesca 2023 ELETTRONICO https://hdl.handle.net/11582/336789 https://doi.org/10.1109/TGRS.2023.3252939 https://ieeexplore.ieee.org/document/10059004 eng eng volume:61 journal:IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING https://hdl.handle.net/11582/336789 doi:10.1109/TGRS.2023.3252939 https://ieeexplore.ieee.org/document/10059004 info:eu-repo/semantics/restrictedAccess info:eu-repo/semantics/article 2023 ftfbkssleriris https://doi.org/10.1109/TGRS.2023.3252939 2024-08-29T03:34:16Z Airborne radar sounders (RSs) are active sensors that acquire subsurface data for Earth observation. RS data (radargrams) provide information on buried geology by imaging subsurface dielectric discontinuities. Recently, several automatic RS target identification techniques have been proposed, with convolutional neural network (CNN)-based methods being the most promising. However, they require numerous labeled data that are hard to retrieve in the subsurface environment targeted by RS. Furthermore, they are not designed to effectively deal with problems showing unbalanced classes, such as RS segmentation. We introduce newer cryosphere subsurface targets in the inland and coastal areas that can have a very low probability. To deal with the higher complexity and variability than previous works, we propose a transfer learning framework for RS data to mitigate the need for a large amount of labeled data and handle extremely unbalanced target classes. Herewith, we propose two transfer learning-based mechanisms for radargram segmentation. The first uses a lightweight architecture whose pretraining is supervised with a large labeled dataset from other domains. The second mechanism uses a deep architecture pretrained in the RS domain, considering the pretest task of radargram reconstruction. The architectures are modified to deal with the characteristics of RS data and the radargram segmentation task. Finally, both methods are fine-tuned with a few labeled radargrams to learn radargram features useful for segmentation. We reveal experimental results on radargrams acquired in Antarctica by MCoRDS-1 and MCoRDS-3. The results demonstrate the effectiveness of transfer learning for radargram segmentation. Article in Journal/Newspaper Antarc* Antarctica Fondazione Bruno Kessler: CINECA IRIS IEEE Transactions on Geoscience and Remote Sensing 61 1 18 |
institution |
Open Polar |
collection |
Fondazione Bruno Kessler: CINECA IRIS |
op_collection_id |
ftfbkssleriris |
language |
English |
description |
Airborne radar sounders (RSs) are active sensors that acquire subsurface data for Earth observation. RS data (radargrams) provide information on buried geology by imaging subsurface dielectric discontinuities. Recently, several automatic RS target identification techniques have been proposed, with convolutional neural network (CNN)-based methods being the most promising. However, they require numerous labeled data that are hard to retrieve in the subsurface environment targeted by RS. Furthermore, they are not designed to effectively deal with problems showing unbalanced classes, such as RS segmentation. We introduce newer cryosphere subsurface targets in the inland and coastal areas that can have a very low probability. To deal with the higher complexity and variability than previous works, we propose a transfer learning framework for RS data to mitigate the need for a large amount of labeled data and handle extremely unbalanced target classes. Herewith, we propose two transfer learning-based mechanisms for radargram segmentation. The first uses a lightweight architecture whose pretraining is supervised with a large labeled dataset from other domains. The second mechanism uses a deep architecture pretrained in the RS domain, considering the pretest task of radargram reconstruction. The architectures are modified to deal with the characteristics of RS data and the radargram segmentation task. Finally, both methods are fine-tuned with a few labeled radargrams to learn radargram features useful for segmentation. We reveal experimental results on radargrams acquired in Antarctica by MCoRDS-1 and MCoRDS-3. The results demonstrate the effectiveness of transfer learning for radargram segmentation. |
author2 |
Hoyo Garcia, Miguel Donini, Elena Bovolo, Francesca |
format |
Article in Journal/Newspaper |
author |
Miguel Hoyo Garcia Elena Donini Francesca Bovolo |
spellingShingle |
Miguel Hoyo Garcia Elena Donini Francesca Bovolo A weakly supervised transfer learning approach for radar sounder data segmentation |
author_facet |
Miguel Hoyo Garcia Elena Donini Francesca Bovolo |
author_sort |
Miguel Hoyo Garcia |
title |
A weakly supervised transfer learning approach for radar sounder data segmentation |
title_short |
A weakly supervised transfer learning approach for radar sounder data segmentation |
title_full |
A weakly supervised transfer learning approach for radar sounder data segmentation |
title_fullStr |
A weakly supervised transfer learning approach for radar sounder data segmentation |
title_full_unstemmed |
A weakly supervised transfer learning approach for radar sounder data segmentation |
title_sort |
weakly supervised transfer learning approach for radar sounder data segmentation |
publishDate |
2023 |
url |
https://hdl.handle.net/11582/336789 https://doi.org/10.1109/TGRS.2023.3252939 https://ieeexplore.ieee.org/document/10059004 |
genre |
Antarc* Antarctica |
genre_facet |
Antarc* Antarctica |
op_relation |
volume:61 journal:IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING https://hdl.handle.net/11582/336789 doi:10.1109/TGRS.2023.3252939 https://ieeexplore.ieee.org/document/10059004 |
op_rights |
info:eu-repo/semantics/restrictedAccess |
op_doi |
https://doi.org/10.1109/TGRS.2023.3252939 |
container_title |
IEEE Transactions on Geoscience and Remote Sensing |
container_volume |
61 |
container_start_page |
1 |
op_container_end_page |
18 |
_version_ |
1810492148084113408 |