EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Advancing SAR Target Recognition Through Hierarchical Self-Supervised Learning with Multi-Task Pretext Training

Md Al Siam iD, Dewan Fahim Noor, Mandoye Ndoye, Jesmin Farzana Khan

Sensors · 24.12.2025 · 26 (1) · 122

DOI: 10.3390/s26010122 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 24.12.2025. Verfasst von Md Al Siam und 3 weiteren Autor:innen. Beteiligte Einrichtung: Electrical & Computer Engineering Department, Tuskegee University, Tuskegee, AL 36088, USA. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 42 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: U.S. National Science Foundation, National Science Foundation and by DoD OUSD (R&E) under Cooperative Agreement.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Synthetic Aperture Radar (SAR) Automatic Target Recognition (ATR) systems face significant challenges due to limited labeled data availability and persistent domain gaps between synthetic and measured imagery. This paper presents a comprehensive self-supervised learning (SSL) framework that eliminates dependency on synthetic data while achieving state-of-the-art performance through multi-task pretext training and extensive downstream classifier evaluation. We systematically evaluate our SSL framework across diverse downstream classifiers spanning different computational paradigms and architectural families. Our study encompasses traditional machine learning approaches (SVM, Random Forest, XGBoost, Gradient Boosting), deep convolutional neural networks (ResNet, U-Net, MobileNet, EfficientNet), and a generative adversarial network. We conduct extensive experiments using the SAMPLE dataset with rigorous evaluation protocols. Results demonstrate that SSL significantly improves SAR ATR performance, with SVM achieving 99.63% accuracy, ResNet18 reaching 97.40% accuracy, and Random Forest demonstrating 99.26% accuracy. Our multi-task SSL framework employs nine carefully designed pretext tasks, including geometric invariance, signal robustness, and multi-scale analysis. Cross-validation experiments validate the generalizability and robustness of our findings. Rigorous comparison with SimCLR baseline validates that task-based SSL outperforms contrastive learning for SAR ATR. This work establishes a new paradigm for SAR ATR that leverages inherent radar data structure without synthetic augmentation, providing practical guidelines for deploying SSL-based SAR ATR systems and a foundation for future domain-specific self-supervised learning research in remote sensing applications.

Schlagwörter

Stichwörter der Autor:innen

automatic target recognition (ATR)computer visiondeep learningmachine learningremote sensingself-supervised learning (SSL)signal processingsynthetic aperture radar (SAR)

Autor:innen und Institutionen

  • Electrical & Computer Engineering Department, Tuskegee University, Tuskegee, AL 36088, USA.

ORCID-Profile: Md Al Siam

Förderung

  • U.S. National Science Foundation
  • National Science Foundation and by DoD OUSD (R&E) under Cooperative Agreement

Zitieren

APA

Siam, M. A., Noor, D. F., Ndoye, M., & Khan, J. F. (2025). Advancing SAR Target Recognition Through Hierarchical Self-Supervised Learning with Multi-Task Pretext Training. Sensors, 26(1), 122. https://doi.org/10.3390/s26010122

Vancouver

Siam MA, Noor DF, Ndoye M, Khan JF. Advancing SAR Target Recognition Through Hierarchical Self-Supervised Learning with Multi-Task Pretext Training. Sensors. 2025;26(1):122. doi:10.3390/s26010122

BibTeX
@article{Siam2025,
  author = {Siam, Md Al and Noor, Dewan Fahim and Ndoye, Mandoye and Khan, Jesmin Farzana},
  title = {Advancing SAR Target Recognition Through Hierarchical Self-Supervised Learning with Multi-Task Pretext Training},
  journal = {Sensors},
  year = {2025},
  volume = {26},
  number = {1},
  pages = {122},
  doi = {10.3390/s26010122}
}
Kostenlos starten →