@article{ART003383487},
author={Ju-Mi Park and Da-Huin Moon and Hyeon-Mo Kim and Woo-Hyeon Moon and Seo-Ho Lee and Hyo-Jo Lee and Won-Seok Jang and Ji-Seok Yoon},
title={Label-Efficient Maritime Radar Target Detection for Intelligent Combat Management System via Domain Adapted Self-Supervised Learning},
journal={Journal of The Korea Society of Computer and Information},
issn={1598-849X},
year={2026},
volume={31},
number={9},
pages={57-76}
TY - JOUR
AU - Ju-Mi Park
AU - Da-Huin Moon
AU - Hyeon-Mo Kim
AU - Woo-Hyeon Moon
AU - Seo-Ho Lee
AU - Hyo-Jo Lee
AU - Won-Seok Jang
AU - Ji-Seok Yoon
TI - Label-Efficient Maritime Radar Target Detection for Intelligent Combat Management System via Domain Adapted Self-Supervised Learning
JO - Journal of The Korea Society of Computer and Information
PY - 2026
VL - 31
IS - 9
PB - The Korean Society Of Computer And Information
SP - 57
EP - 76
SN - 1598-849X
AB - Maritime radar provides essential wide-area surveillance for naval Combat Management System (CMS), yet 'object-like' clutter often leads to False Positives (FPs) that compromise system reliability.
This research focuses on analyzing these FPs to develop strategies for enhancing the autonomy and reliability of maritime situational awareness. First, we empirically establish the operational performance ceiling for post-detection FP suppression (AUROC ≈ 0.844), demonstrating that hand-crafted features are fundamentally insufficient to resolve the physical indistinguishability of clutter. To overcome this limitation, we propose a Radar-Aware MAE (RA-MAE), a label-efficient detection framework based on a domain-adapted self-supervised learning strategy. In particular, a Blob-Priority Masking module is introduced to encourage the encoder to extract high-dimensional structural patterns from extensive unlabeled radar data. Experimental results demonstrate that RA-MAE significantly improves detection performance in data-scarce regimes. Notably, with a 10% label ratio, the proposed framework achieved a 9.2% reduction in FPs while maintaining target recall. Furthermore, a comprehensive analysis provides practical operational guidelines to optimize the trade-off between annotation costs and computational resource investment. Consequently, this research establishes a viable technical trajectory for enhancing the operational reliability and autonomy of naval maritime situational awareness by transitioning from heuristic feature engineering to representation learning.
KW - Intelligent Combat Management System;False Positive Suppression;Self-Supervised Learning;Maritime Radar Representation Learning;Label-Efficient Learning
DO -
UR -
ER -
Ju-Mi Park, Da-Huin Moon, Hyeon-Mo Kim, Woo-Hyeon Moon, Seo-Ho Lee, Hyo-Jo Lee, Won-Seok Jang and Ji-Seok Yoon. (2026). Label-Efficient Maritime Radar Target Detection for Intelligent Combat Management System via Domain Adapted Self-Supervised Learning. Journal of The Korea Society of Computer and Information, 31(9), 57-76.
Ju-Mi Park, Da-Huin Moon, Hyeon-Mo Kim, Woo-Hyeon Moon, Seo-Ho Lee, Hyo-Jo Lee, Won-Seok Jang and Ji-Seok Yoon. 2026, "Label-Efficient Maritime Radar Target Detection for Intelligent Combat Management System via Domain Adapted Self-Supervised Learning", Journal of The Korea Society of Computer and Information, vol.31, no.9 pp.57-76.
Ju-Mi Park, Da-Huin Moon, Hyeon-Mo Kim, Woo-Hyeon Moon, Seo-Ho Lee, Hyo-Jo Lee, Won-Seok Jang, Ji-Seok Yoon "Label-Efficient Maritime Radar Target Detection for Intelligent Combat Management System via Domain Adapted Self-Supervised Learning" Journal of The Korea Society of Computer and Information 31.9 pp.57-76 (2026) : 57.
Ju-Mi Park, Da-Huin Moon, Hyeon-Mo Kim, Woo-Hyeon Moon, Seo-Ho Lee, Hyo-Jo Lee, Won-Seok Jang, Ji-Seok Yoon. Label-Efficient Maritime Radar Target Detection for Intelligent Combat Management System via Domain Adapted Self-Supervised Learning. 2026; 31(9), 57-76.
Ju-Mi Park, Da-Huin Moon, Hyeon-Mo Kim, Woo-Hyeon Moon, Seo-Ho Lee, Hyo-Jo Lee, Won-Seok Jang and Ji-Seok Yoon. "Label-Efficient Maritime Radar Target Detection for Intelligent Combat Management System via Domain Adapted Self-Supervised Learning" Journal of The Korea Society of Computer and Information 31, no.9 (2026) : 57-76.
Ju-Mi Park; Da-Huin Moon; Hyeon-Mo Kim; Woo-Hyeon Moon; Seo-Ho Lee; Hyo-Jo Lee; Won-Seok Jang; Ji-Seok Yoon. Label-Efficient Maritime Radar Target Detection for Intelligent Combat Management System via Domain Adapted Self-Supervised Learning. Journal of The Korea Society of Computer and Information, 31(9), 57-76.
Ju-Mi Park; Da-Huin Moon; Hyeon-Mo Kim; Woo-Hyeon Moon; Seo-Ho Lee; Hyo-Jo Lee; Won-Seok Jang; Ji-Seok Yoon. Label-Efficient Maritime Radar Target Detection for Intelligent Combat Management System via Domain Adapted Self-Supervised Learning. Journal of The Korea Society of Computer and Information. 2026; 31(9) 57-76.
Ju-Mi Park, Da-Huin Moon, Hyeon-Mo Kim, Woo-Hyeon Moon, Seo-Ho Lee, Hyo-Jo Lee, Won-Seok Jang, Ji-Seok Yoon. Label-Efficient Maritime Radar Target Detection for Intelligent Combat Management System via Domain Adapted Self-Supervised Learning. 2026; 31(9), 57-76.
Ju-Mi Park, Da-Huin Moon, Hyeon-Mo Kim, Woo-Hyeon Moon, Seo-Ho Lee, Hyo-Jo Lee, Won-Seok Jang and Ji-Seok Yoon. "Label-Efficient Maritime Radar Target Detection for Intelligent Combat Management System via Domain Adapted Self-Supervised Learning" Journal of The Korea Society of Computer and Information 31, no.9 (2026) : 57-76.
