AI기반 수중기뢰 탐지를 위한 AI 모델 경량화 기법
Ⓒ 2026 Korea Society for Naval Science & Technology
초록
수중기뢰탐지는 해양 안전확보와 군사작전에서 핵심적인 임무로, 최근에는 수중 소나 영상 데이터를 딥러닝 기법으로 분석하여 기뢰 탐지를 수행하는 연구가 활발히 진행되고 있다. 특히 자율무인잠수정 및 무인수상정에 이러한 모델을 실시간으로 탑재-운용하기 위해서는 제한된 연산자원과 전력환경에 적합한 경량화 기법이 필수적이다. 본 논문은 먼저 수중기뢰 탐지 분야에서의 딥러닝 활용 동향을 개관하고, 대표적인 경량화 기법들을 소개한다. 또한 소나 영상 기반 탐지에 적용된 사례들을 정리하고, 성능, 효율 개선 효과를 분석한다. 마지막으로, 이러한 기법들의 실전 운용시 기대효과와 향후 연구방향을 제안한다.
Abstract
Underwater mine detection is vital for maritime safety and naval operations. Deep learning has recently been applied to sonar imagery for this task. However, real-time deployment on Automated Underwater Vehicles (AUVs) and Unmanned Surface Vehicles (USVs) requires lightweight and efficient models. This paper reviews current deep learning-based sonar detection methods, outlines key model compression techniques, and summarizes representative applications. We also examine their impact on accuracy and efficiency and discuss implications for field deployment and future research.
Keywords:
Deep Learning, Quantization)모델 경량화(Model Compression, Detection)수중기뢰탐지(Underwater Mine Detection키워드:
딥러닝, 양자화, 탐지Acknowledgments
이 논문은 2023년도 정부(방위사업청)의 재원으로 국방기술진흥연구소의 지원을 받아 수행된 연구임(No. KRIT-CT-23-035-03, AI기반 수중 기뢰 탐지 기술(기뢰탐지용 무인잠수정 군집 운용 기술)).
References
- Sifan Zhou, Liang Li, Xinyu Zhang, Bo Zhang, Shipeng Bai, Miao Sun, Ziyu Zhao, Xiaobo Lu, & Xiangxiang Chu, ‘LiDAR-PTQ: Post-Training Quantization for Point Cloud 3D Object Detection,’ arXiv preprint arXiv:2401.15865, , 2024.
- Abhinav Goel, Caleb Tung, Yung-Hsiang Lu, & George K. Thiruvathukal, ‘A Survey of Methods for Low-Power Deep Learning and Computer Vision,’ in proceedings of 2020 IEEE 6th World Forum on Internet of Things (WF-IoT), New Orleans, LA, 2020.
- Zhuang Liu, Jianguo Li, Zhiqiang Shen, Gao Huang, Shoumeng Yan, Changshui Zhang, ‘Learning Efficient Convolutional Networks Through Network Slimming,’ Proceedings of the IEEE International Conference On Computer Vision (pp. 2736-2744), 2017.
- Minjie Y. Wang, ‘Deep Graph Library: Towards Efficient and Scalable Deep Learning on Graphs.’ ICLR Workshop on Representation Learning on Graphs and Manifolds, 2019.
- Martin Aubard, László Antal, Ana Madureira, & Erika Ábrahám, ‘Knowledge Distillation in YOLOX-Vit for Side-Scan Sonar Object Detection,’ arxiv preprint arXiv:2403.09313, , 2024.
-
Xinwei Ou, Zhangxin Chen, Ce Zhu, & Yipeng Liu, ‘Low Rank Optimization for Efficient Deep Learning: Making a Balance Between Compact Architecture and Fast Training,’ Journal of Systems Engineering and Electronics, VOL. 35, NO. 3, 2023, pp. 509-531.
[https://doi.org/10.23919/JSEE.2023.000159]
-
Tara N. Sainath, Brian Kingsbury, Vikas Sindhwani, Ebru Arisoy, & Bhuvana Ramabhadran, ‘Low-Rank Matrix Factorization for Deep Neural Network Training with High-Dimensional Output Targets,’ in proceedings of 2013 IEEE International Conference on Acoustics, Speech and Signal Processing, Vancouver, Canada, 2013.
[https://doi.org/10.1109/ICASSP.2013.6638949]
-
Andrew Howard, Mark Sandler, Grace Chu, Liang-Chieh Chen, Bo Chen, Mingxing Tan, Weijun Wang, Yukun Zhu, Ruoming Pang, & Vijay Vasudevan, ‘Searching for MobileNetV3,’ Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 1314-1324), 2019.
[https://doi.org/10.1109/ICCV.2019.00140]
-
Chensheng Cheng, Xujia Hou, Can Wang, Xin Wen, Weidong Liu, & Feihu Zhang, ‘A Pruning and Distillation Based Compression Method for Sonar Image Detection Models,’ Journal of Marine Science and Engineering, VOL. 12, NO. 6, 2024, 1033.
[https://doi.org/10.3390/jmse12061033]
-
Yibing Fan, Lanyong Zhang, & Peng Li, ‘A Lightweight Model of Underwater Object Detection Based on YOLOv8n for an Edge Computing Platform,’ Journal of Marine Science and Engineering, VOL. 12, NO. 5, 2024, 697.
[https://doi.org/10.3390/jmse12050697]
- Lin Niu, Jiawei Liu, Zhihang Yuan, Dawei Yang, Xinggang Wang, & Wenyu Liu, ‘Improving Post-Training Quantization on Object Detection with Task Loss-Guided Lp Metric,’ arxiv preprint arXiv:2304.09785, , 2023.