Unsupervised Domain Adaptative SAR Target Detection Based on Feature Decomposition and Uncertainty-Guided Self-Training

This article proposes an unsupervised domain adaptation (UDA) method by transferring knowledge from rich labeled optical domain to unlabeled synthetic aperture radar (SAR) domain, tackling the issue that current deep-learning-based SAR target detection methods rely on abundant labeled SAR images. Sp...

Full description

Saved in:
Bibliographic Details
Main Authors: Yu Shi, Yi Li, Lan Du, Yuang Du, Yuchen Guo
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/10750353/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1846163870757945344
author Yu Shi
Yi Li
Lan Du
Yuang Du
Yuchen Guo
author_facet Yu Shi
Yi Li
Lan Du
Yuang Du
Yuchen Guo
author_sort Yu Shi
collection DOAJ
description This article proposes an unsupervised domain adaptation (UDA) method by transferring knowledge from rich labeled optical domain to unlabeled synthetic aperture radar (SAR) domain, tackling the issue that current deep-learning-based SAR target detection methods rely on abundant labeled SAR images. Specifically, we gradually encode the dependencies across different granularity perspectives including domain invariant representations (DIR) learning based on feature decomposition and domain discriminative representations (DDR) learning based on uncertainty-guided self-training. First, existing methods usually learn the DIR by directly minimizing domain discrepancy between two domains, which is difficult to achieve in practice. Due to the huge difference between the optical and SAR images, rich domain-specific characteristics bring great challenges to learn the DIR. To alleviate the above difficulty, we explicitly model the domain-invariant and domain-specific features in the representations by constructing a network with feature decomposition to better extract the DIR across domains, where only the DIR extracted from optical images and their labels are used to train the domain-shared detector in this stage. Second, even DIR can be extracted, the domain-shared detector will lose some discriminative and valuable features of the SAR domain while minimizing the distribution discrepancy between the SAR and labeled optical domain. In order to achieve the better detection performance for SAR images, a self-training method based on pseudolabels is proposed to learn DDR and train the SAR-dedicated detector. Furthermore, for ensuring the reliability of pseudolabels, we present a novel uncertainty-guided pseudolabel selection strategy, which contains two phases: one is instance uncertainty guided selection, the other is image uncertainty guided selection. Finally, based on measured optical and SAR datasets, we conduct extensive empirical evaluation to verify the effectuality of our proposed method.
format Article
id doaj-art-df64089c38624c62b8063cfa99d5e86a
institution Kabale University
issn 1939-1404
2151-1535
language English
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-df64089c38624c62b8063cfa99d5e86a2024-11-19T00:00:53ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352024-01-0117202652028310.1109/JSTARS.2024.348692210750353Unsupervised Domain Adaptative SAR Target Detection Based on Feature Decomposition and Uncertainty-Guided Self-TrainingYu Shi0https://orcid.org/0000-0002-5081-231XYi Li1https://orcid.org/0000-0002-8192-308XLan Du2https://orcid.org/0000-0002-4503-0022Yuang Du3https://orcid.org/0000-0002-6525-7056Yuchen Guo4https://orcid.org/0000-0003-3424-7231National Key Laboratory of Radar Signal Processing, Xidian University, Xi'an, ChinaNational Key Laboratory of Radar Signal Processing, Xidian University, Xi'an, ChinaNational Key Laboratory of Radar Signal Processing, Xidian University, Xi'an, ChinaNational Key Laboratory of Radar Signal Processing, Xidian University, Xi'an, ChinaAcademy of Advanced Interdisciplinary Research, Xidian University, Xi'an, ChinaThis article proposes an unsupervised domain adaptation (UDA) method by transferring knowledge from rich labeled optical domain to unlabeled synthetic aperture radar (SAR) domain, tackling the issue that current deep-learning-based SAR target detection methods rely on abundant labeled SAR images. Specifically, we gradually encode the dependencies across different granularity perspectives including domain invariant representations (DIR) learning based on feature decomposition and domain discriminative representations (DDR) learning based on uncertainty-guided self-training. First, existing methods usually learn the DIR by directly minimizing domain discrepancy between two domains, which is difficult to achieve in practice. Due to the huge difference between the optical and SAR images, rich domain-specific characteristics bring great challenges to learn the DIR. To alleviate the above difficulty, we explicitly model the domain-invariant and domain-specific features in the representations by constructing a network with feature decomposition to better extract the DIR across domains, where only the DIR extracted from optical images and their labels are used to train the domain-shared detector in this stage. Second, even DIR can be extracted, the domain-shared detector will lose some discriminative and valuable features of the SAR domain while minimizing the distribution discrepancy between the SAR and labeled optical domain. In order to achieve the better detection performance for SAR images, a self-training method based on pseudolabels is proposed to learn DDR and train the SAR-dedicated detector. Furthermore, for ensuring the reliability of pseudolabels, we present a novel uncertainty-guided pseudolabel selection strategy, which contains two phases: one is instance uncertainty guided selection, the other is image uncertainty guided selection. Finally, based on measured optical and SAR datasets, we conduct extensive empirical evaluation to verify the effectuality of our proposed method.https://ieeexplore.ieee.org/document/10750353/Feature decompositionoptical to synthetic aperture radar (SAR)SARtarget detectionuncertainty-guided self-trainingunsupervised domain adaptation
spellingShingle Yu Shi
Yi Li
Lan Du
Yuang Du
Yuchen Guo
Unsupervised Domain Adaptative SAR Target Detection Based on Feature Decomposition and Uncertainty-Guided Self-Training
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Feature decomposition
optical to synthetic aperture radar (SAR)
SAR
target detection
uncertainty-guided self-training
unsupervised domain adaptation
title Unsupervised Domain Adaptative SAR Target Detection Based on Feature Decomposition and Uncertainty-Guided Self-Training
title_full Unsupervised Domain Adaptative SAR Target Detection Based on Feature Decomposition and Uncertainty-Guided Self-Training
title_fullStr Unsupervised Domain Adaptative SAR Target Detection Based on Feature Decomposition and Uncertainty-Guided Self-Training
title_full_unstemmed Unsupervised Domain Adaptative SAR Target Detection Based on Feature Decomposition and Uncertainty-Guided Self-Training
title_short Unsupervised Domain Adaptative SAR Target Detection Based on Feature Decomposition and Uncertainty-Guided Self-Training
title_sort unsupervised domain adaptative sar target detection based on feature decomposition and uncertainty guided self training
topic Feature decomposition
optical to synthetic aperture radar (SAR)
SAR
target detection
uncertainty-guided self-training
unsupervised domain adaptation
url https://ieeexplore.ieee.org/document/10750353/
work_keys_str_mv AT yushi unsuperviseddomainadaptativesartargetdetectionbasedonfeaturedecompositionanduncertaintyguidedselftraining
AT yili unsuperviseddomainadaptativesartargetdetectionbasedonfeaturedecompositionanduncertaintyguidedselftraining
AT landu unsuperviseddomainadaptativesartargetdetectionbasedonfeaturedecompositionanduncertaintyguidedselftraining
AT yuangdu unsuperviseddomainadaptativesartargetdetectionbasedonfeaturedecompositionanduncertaintyguidedselftraining
AT yuchenguo unsuperviseddomainadaptativesartargetdetectionbasedonfeaturedecompositionanduncertaintyguidedselftraining