온-보드 항공 표적 탐지 모델 성능 개선: 형상 정보와 혼합 정밀도 연산의 활용
Improving On-board Aircraft Target Detection Model: Utilizing Shape Information and Mixed Precision Quantization
1
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
항공 표적을 전자 광학 시스템으로 탐지하는 데에는 컴퓨터 비전 분야의 객체 탐지 기술이 이용된다. 최신의 객체 탐지 기술은 심층 신경망 기술과 이를 가속하는 하드웨어의 발전으로 성능이 비약적으로 향상되었는데, 이는 항공 표적 도메인에도 적용될 수 있다. 그러나 발전된 기술을실제 무기 체계에 적용하기 위해서는 온-보드(On-board) 환경에서 높은 탐지 신뢰도와 실시간성을달성할 수 있어야 한다. 본 논문에서는 이에 적합한 방법들을 두 가지 전략으로 구분하여 소개한다. 전략은 객체 형상 정보를 이용한 모델 훈련 개선과 복합 정밀도 연산 기반의 최적화로 요약될 수 있다. 실험 결과는 논문의 방법이 항공 표적 탐지 모델의 성능을 개선하고 무기 체계 적용성을 향상시킬 수 있음을 보여준다.
Abstract
The object detection technology in the field of computer vision is utilized to detect aircraft targets using electro-optical systems. Recent advancements in object detection, driven by deep neural networks and hardware accelerators, have led to significant performance improvements, which can also be applied to aircraft target detection. However, for successful integration into weapon systems, it is essential to achieve high detection reliability and real-time processing in on-board environments. This paper presents two strategies to address these challenges. These strategies focus on improving model training by utilizing object shape information and optimizing the model with mixed precision computation. The experimental results demonstrate that the methods improve the performance of the aircraft target detection model and enhance its applicability to weapon systems.
참고문헌 (24)
- R. Girshick, "Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.580~587 (2014)
- R. Girshick, "Fast R-CNN", Proceedings of the IEEE International Conference on Computer Vision, pp.1440~1448 (2015)
- S. Ren, "Faster R-CNN : Towards Real-Time Object Detection with Region Proposal Networks", IEEE Transactions on Pattern Analysis and Machine Intelligence, 39(6), pp.1137~1149 (2017)[DOI]
- J. Redmon, "You Only Look Once : Unified, Real-Time Object Detection", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.779~788 (2016)
- "Ultralytics"[URL]
- Z. Tian, "FCOS : Fully Convolutional One-Stage Object Detection", Proceedings of the IEEE, 2019, 9627~9636
- X. Hou, "Airborne Infrared Aircraft Target Detection Algorithm Based on YOLOv4-tiny", Proceedings of the 2021 International Conference on Advances in Optics and Computational Sciences (2021)
- Y. Liu, "Research on Infrared Dim Target Detection Based on Improved YOLOv8", Remote Sensing, 16(16) (2024)[DOI]
- "Hailo-8 NPU"[URL]
- D. -H. Hur, "Object Detection Performance Analysis between On-GPU and On-Board Analysis for Military Domain Images", Journal of the Korea Society of Computer and Information, 29(8), pp.157~164 (2024)[DOI]
- Challenges and Implications of Testing and Evaluation of AI Autonomous Weapon Systems, "H. Jung", 24~32, pp.KIDA Defense Issues & Analyses, No. 2005 (2024)
- P. Zhang, "SlimYOLOv3 : Narrower, Faster and Better for Real-Time UAV Applications", IEEE, 2019, 37~45
- G. Chen, "Learning Efficient Object Detection Models with Knowledge Distillation", Proceedings of the 31st International Conference on Neural Information Processing Systems, pp.742~751 (2017)
- P. Pérez, "Poisson Image Editing", ACM Transactions on Graphics, 22(3), pp.313~318 (2003)[DOI]
- "FLIR A6701"[URL]
- "AnyLabeling"[URL]
- "COCO"[URL]
- "Raspberry Pi 5"[URL]
- B. Jacob, "Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.2704~2713 (2018)