Automated Ship Detection in Remote Sensing Imagery using Knowledge Distillation Feature Maps
Defence Science Journal · 29.04.2026 · 76 (3) · 390-398
DOI: 10.14429/dsj.21624 ↗Kurz gefasst
Fachartikel in der Zeitschrift Defence Science Journal, online erschienen am 29.04.2026. Verfasst von Ghanshyam Meena und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 2.5).
Abstract (Originalsprache)
Automated ship detection and subsequent classification of maritime vessels operating within littoral zones are paramount for effectively safeguarding national and international maritime interests regarding ship traffic. Over recent years, different deep learning approaches have made notable progress in image analysis and classification. Still, it requires high processor power and memory space, which poses deployment challenges for such a system on satellites and edge devices. To address this, a lightweight deep learning network is designed to minimize computational parameters and complexity while retaining high recognition rate in ship classification task. The proposed method has incorporated the knowledge distillation technique, where a high-capacity teacher network guides the lightweight student model by transferring soft output distributions. Experimental results affirm that the proposed lightweight method has shown superior ship image classification performance compared to existing approaches, achieving the highest classification accuracy of 97.25% with least 2.35M parameters.
Zitieren
APA
Meena, G., Gupta, S., & Tripathi, A. K. (2026). Automated Ship Detection in Remote Sensing Imagery using Knowledge Distillation Feature Maps. Defence Science Journal, 76(3), 390-398. https://doi.org/10.14429/dsj.21624
Vancouver
Meena G, Gupta S, Tripathi AK. Automated Ship Detection in Remote Sensing Imagery using Knowledge Distillation Feature Maps. Defence Science Journal. 2026;76(3):390-398. doi:10.14429/dsj.21624
BibTeX
@article{Meena2026,
author = {Meena, Ghanshyam and Gupta, Sachin and Tripathi, Ashish Kumar},
title = {Automated Ship Detection in Remote Sensing Imagery using Knowledge Distillation Feature Maps},
journal = {Defence Science Journal},
year = {2026},
volume = {76},
number = {3},
pages = {390--398},
doi = {10.14429/dsj.21624}
}
EUVIMED