Multimodal Colearning Meets Remote Sensing: Taxonomy, State of the Art, and Future Works
In remote sensing (RS), multiple modalities of data are usually available, e.g., RGB, multispectral, hyperspectral, light detection and ranging (LiDAR), and synthetic aperture radar (SAR). Multimodal machine learning systems, which fuse these rich multimodal data modalities, have shown better perfor...
Main Authors: | , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2024-01-01
|
Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/10474099/ |
_version_ | 1797217407569231872 |
---|---|
author | Nhi Kieu Kien Nguyen Abdullah Nazib Tharindu Fernando Clinton Fookes Sridha Sridharan |
author_facet | Nhi Kieu Kien Nguyen Abdullah Nazib Tharindu Fernando Clinton Fookes Sridha Sridharan |
author_sort | Nhi Kieu |
collection | DOAJ |
description | In remote sensing (RS), multiple modalities of data are usually available, e.g., RGB, multispectral, hyperspectral, light detection and ranging (LiDAR), and synthetic aperture radar (SAR). Multimodal machine learning systems, which fuse these rich multimodal data modalities, have shown better performance compared to unimodal systems. Most multimodal research assumes that all modalities are present, aligned, and noiseless during training and testing time. However, in real-world scenarios, it is common to observe that one or more modalities are missing, noisy, and nonaligned, in either training or testing or both. In addition, acquiring large-scale, noise-free annotations is expensive, as a result, lacking sufficient annotated datasets or having to deal with inconsistent labels are open challenges. These challenges can be addressed under a learning paradigm called multimodal colearning. This article focuses on multimodal colearning techniques for RS data. We first review what data modalities are available in the RS domain and the key benefits and challenges of combining multimodal data in the RS context. We then review the RS tasks that would benefit from multimodal processing including classification, segmentation, target detection, anomaly detection, and temporal change detection. We then dive deeper into technical details by reviewing more than 200 recent efforts in this area and provide a comprehensive taxonomy to systematically review state-of-the-art approaches in four key colearning challenges including missing modalities, noisy modalities, limited modality annotations, and weakly paired modalities. Based on these insights, we propose emerging research directions to inform potential future research in multimodal colearning for RS. |
first_indexed | 2024-04-24T12:01:22Z |
format | Article |
id | doaj.art-e877b3ac8e8a430785d4a55e4adc2beb |
institution | Directory Open Access Journal |
issn | 2151-1535 |
language | English |
last_indexed | 2024-04-24T12:01:22Z |
publishDate | 2024-01-01 |
publisher | IEEE |
record_format | Article |
series | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
spelling | doaj.art-e877b3ac8e8a430785d4a55e4adc2beb2024-04-08T23:00:12ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing2151-15352024-01-01177386740910.1109/JSTARS.2024.337834810474099Multimodal Colearning Meets Remote Sensing: Taxonomy, State of the Art, and Future WorksNhi Kieu0Kien Nguyen1https://orcid.org/0000-0002-3466-9218Abdullah Nazib2https://orcid.org/0000-0003-1048-0346Tharindu Fernando3https://orcid.org/0000-0002-6935-1816Clinton Fookes4https://orcid.org/0000-0002-8515-6324Sridha Sridharan5https://orcid.org/0000-0003-4316-9001School of Electrical Engineering and Robotics, Queensland University of Technology, Brisbane, QLD, AustraliaSchool of Electrical Engineering and Robotics, Queensland University of Technology, Brisbane, QLD, AustraliaSchool of Electrical Engineering and Robotics, Queensland University of Technology, Brisbane, QLD, AustraliaSchool of Electrical Engineering and Robotics, Queensland University of Technology, Brisbane, QLD, AustraliaSchool of Electrical Engineering and Robotics, Queensland University of Technology, Brisbane, QLD, AustraliaSchool of Electrical Engineering and Robotics, Queensland University of Technology, Brisbane, QLD, AustraliaIn remote sensing (RS), multiple modalities of data are usually available, e.g., RGB, multispectral, hyperspectral, light detection and ranging (LiDAR), and synthetic aperture radar (SAR). Multimodal machine learning systems, which fuse these rich multimodal data modalities, have shown better performance compared to unimodal systems. Most multimodal research assumes that all modalities are present, aligned, and noiseless during training and testing time. However, in real-world scenarios, it is common to observe that one or more modalities are missing, noisy, and nonaligned, in either training or testing or both. In addition, acquiring large-scale, noise-free annotations is expensive, as a result, lacking sufficient annotated datasets or having to deal with inconsistent labels are open challenges. These challenges can be addressed under a learning paradigm called multimodal colearning. This article focuses on multimodal colearning techniques for RS data. We first review what data modalities are available in the RS domain and the key benefits and challenges of combining multimodal data in the RS context. We then review the RS tasks that would benefit from multimodal processing including classification, segmentation, target detection, anomaly detection, and temporal change detection. We then dive deeper into technical details by reviewing more than 200 recent efforts in this area and provide a comprehensive taxonomy to systematically review state-of-the-art approaches in four key colearning challenges including missing modalities, noisy modalities, limited modality annotations, and weakly paired modalities. Based on these insights, we propose emerging research directions to inform potential future research in multimodal colearning for RS.https://ieeexplore.ieee.org/document/10474099/Multimodal colearningmultimodal learningremote sensing (RS)satellite imagery |
spellingShingle | Nhi Kieu Kien Nguyen Abdullah Nazib Tharindu Fernando Clinton Fookes Sridha Sridharan Multimodal Colearning Meets Remote Sensing: Taxonomy, State of the Art, and Future Works IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Multimodal colearning multimodal learning remote sensing (RS) satellite imagery |
title | Multimodal Colearning Meets Remote Sensing: Taxonomy, State of the Art, and Future Works |
title_full | Multimodal Colearning Meets Remote Sensing: Taxonomy, State of the Art, and Future Works |
title_fullStr | Multimodal Colearning Meets Remote Sensing: Taxonomy, State of the Art, and Future Works |
title_full_unstemmed | Multimodal Colearning Meets Remote Sensing: Taxonomy, State of the Art, and Future Works |
title_short | Multimodal Colearning Meets Remote Sensing: Taxonomy, State of the Art, and Future Works |
title_sort | multimodal colearning meets remote sensing taxonomy state of the art and future works |
topic | Multimodal colearning multimodal learning remote sensing (RS) satellite imagery |
url | https://ieeexplore.ieee.org/document/10474099/ |
work_keys_str_mv | AT nhikieu multimodalcolearningmeetsremotesensingtaxonomystateoftheartandfutureworks AT kiennguyen multimodalcolearningmeetsremotesensingtaxonomystateoftheartandfutureworks AT abdullahnazib multimodalcolearningmeetsremotesensingtaxonomystateoftheartandfutureworks AT tharindufernando multimodalcolearningmeetsremotesensingtaxonomystateoftheartandfutureworks AT clintonfookes multimodalcolearningmeetsremotesensingtaxonomystateoftheartandfutureworks AT sridhasridharan multimodalcolearningmeetsremotesensingtaxonomystateoftheartandfutureworks |