Journal of the Korea Society of Computer and Information (한국컴퓨터정보학회논문지)

Korean Society of Computer Information (한국컴퓨터정보학회)
권호 표지
DOI QR코드

DOI QR Code

Object Detection Performance Analysis between On-GPU and On-Board Analysis for Military Domain Images

  • Du-Hwan Hur (Vision & Learning Lab, Inha University) ;
  • Dae-Hyeon Park (Vision & Learning Lab, Inha University) ;
  • Deok-Woong Kim (Vision & Learning Lab, Inha University) ;
  • Jae-Yong Baek (Vision & Learning Lab, Inha University) ;
  • Jun-Hyeong Bak (PGM Seeker R&D, LIG Nex1 Co., Ltd.) ;
  • Seung-Hwan Bae (Vision & Learning Lab, Dept. of Electrical and Computer Engineering, Inha University)
  • Received : 2024.06.18
  • Accepted : 2024.07.09
  • Published : 2024.08.30
Download PDF
⟨ Previous Next ⟩

Abstract

In this paper, we propose a discussion that the feasibility of deploying a deep learning-based detector on the resource-limited board. Although many studies evaluate the detector on machines with high-performed GPUs, evaluation on the board with limited computation resources is still insufficient. Therefore, in this work, we implement the deep-learning detectors and deploy them on the compact board by parsing and optimizing a detector. To figure out the performance of deep learning based detectors on limited resources, we monitor the performance of several detectors with different H/W resource. On COCO detection datasets, we compare and analyze the evaluation results of detection model in On-Board and the detection model in On-GPU in terms of several metrics with mAP, power consumption, and execution speed (FPS). To demonstrate the effect of applying our detector for the military area, we evaluate them on our dataset consisting of thermal images considering the flight battle scenarios. As a results, we investigate the strength of deep learning-based on-board detector, and show that deep learning-based vision models can contribute in the flight battle scenarios.

본 논문에서는 제한된 자원을 가진 보드에서 딥러닝 기반 검출기 구축에 대한 실현 가능성에 대해 논의한다. 많은 연구에서 고성능 GPU 환경에서 검출기를 평가하지만, 제한된 연산 자원을 가진 보드에서의 평가는 여전히 미비하다. 따라서 본 연구에서는 검출기를 파싱하고 최적화하는 것으로 보드에 딥러닝 기반 검출기를 구현하고 구축한다. 제한된 자원에서의 딥러닝 기반 검출기의 성능을 확인하기 위해, 여러 검출기를 다양한 하드웨어 자원에서 모니터링하고, COCO 검출 데이터 셋에서 On-Board에서의 검출 모델과 On-GPU의 검출 모델을 mAP, 전력 소모량, 실행 속도(FPS) 관점으로 비교 및 분석한다. 그리고 군사 분야에 검출기를 적용한 효과를 고려하기 위해 항공 전투 시나리오를 고려할 수 있는 열화상 이미지로 구성된 자체 데이터 셋에서 검출기를 평가한다. 결과적으로 우리는 본 연구를 통해 On-Board에서 모델을 실행하는 딥러닝 기반 검출기의 강점을 조사하고, 전장 상황에서 딥러닝 기반 검출기가 기여할 수 있음을 보인다.

Keywords

Acknowledgement

This work was supported by Korea Research In stitute for defense Technology planning and adv ancement(KRIT) grant funded by the Korea gover nment(DAPA(Defense Acquisition Program Admin istration)) (No. 21-107-F00-015(KRIT-CT-22-024-02), 2023)

References

  1. Blalock, Davis, et al. "What is the state of neural network pruning?." Proceedings of the Third Conference on Machine Learning and Systems, pp.129-146, Austin, TX, USA, March 2020. DOI: https://doi.org/10.48550/arXiv.2003.03033
  2. Nagel, Markus, et al. "A white paper on neural network quantization." arXiv preprint arXiv:2106.08295, June 2021. DOI:https://doi.org/10.48550/arXiv.2106.08295
  3. Hinton, Geoffrey, Oriol Vinyals, and Jeff Dean."Distilling the knowledge in a neural network." arXiv preprint arXiv:1503.02531, Aug. 2015. DOI: https://doi.org/10.48550/arXiv.1503.02531
  4. Wang, Xinjiang, Liu, Zeyu, Hu, Yu, Xi, Wei, Yu, Wenxian, & Zou, Danping. "Featurebooster: Boosting feature descriptors with a lightweight neural network." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 7630-7639, Vancouver, BC, Canada, June 2023. DOI: 10.1109/CVPR52729.2023.00737
  5. Cao, Ziang, Chen, Ke, Zhu, Yi, & Zhang, Tao. "Tctrack: Temporal contexts for aerial tracking." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 456-460, LA. USA, June 2022. DOI: 10.1109/CVPR52688.2022.01438
  6. Agrawal, Kshitij, and Anbumani Subramanian. "Enhancing object detection in adverse conditions using thermal imaging." arXiv preprint arXiv:1909.13551, Oct. 2019. DOI: https://doi.org/10.48550/arXiv.1909.13551
  7. Boeing. "F-15K 슬램이글(Slam Eagle)." URL: www.boeing.co.kr/products-and-services/defense-space-and-security/f-15k-slam-eagle.
  8. Ren, Shaoqing, et al. "Faster r-cnn: Towards real-time object detection with region proposal networks." Advances in neural information processing systems 28, pp. 91-99, Quebec, Canada, Dec. 2015. DOI: https://doi.org/10.1109/tpami.2016.2577031
  9. Redmon, Joseph, & Farhadi, Ali. "Yolov3: An incremental improvement." arXiv preprint arXiv:1804.02767, Aug. 2018. DOI:https://doi.org/10.48550/arXiv.1804.02767
  10. Ge, Zheng, et al. "Yolox: Exceeding yolo series in 2021." arXiv preprint arXiv:2107.08430, April 2021. DOI: https://doi.org/10.48550/arXiv.2107.08430
  11. S. Ward-Foxton, "Details of hailo ai edge accelerator emerge,"8 2019.[Online]. URL: https://www.eetimes.com/details-of-hailoai-edge-accelerator-emerge/
  12. Zhang, Hongyi, Cisse, Moustapha, Dauphin, Yann N., & Lopez-Paz, David. "mixup: Beyond empirical risk minimization." arXiv preprint arXiv:1710.09412, Aug. 2017. DOI: https://doi.org/10.48550/arXiv.1710.09412
  13. Netron. "Lutz Roeder's Netron." URL: https://netron.app/
  14. Meller, Eldad, Almog, Uri, & Grobman, Mark. "Same, same but different: Recovering neural network quantization error through weight factorization." International Conference on Machine Learning, PMLR, pp. 4486-4495, Long Beach, California, USA, June 2019. DOI: https://doi.org/10.48550/arXiv.1902.01917
  15. Finkelstein, Alexander, Almog, Uri, & Grobman, Mark. "Fighting quantization bias with bias." arXiv preprint arXiv:1906.03193, June 2019. DOI: https://doi.org/10.48550/arXiv.1906.03193
  16. McKinstry, Jeffrey L., Andersen, David, Wang, Xinjiang, & Chattopadhyay, Aniruddha. "Discovering low-precision networks close to full-precision networks for efficient embedded inference." arXiv preprint arXiv:1809.04191, Oct. 2018. DOI:https://doi.org/10.48550/arXiv.1809.04191
  17. Nagel, Markus, Amjad, Rabab A., Blankevoort, Tijmen, & Welling, Max. "Up or down? adaptive rounding for post-training quantization." International Conference on Machine Learning, PMLR, July 2020. DOI: 10.48550/arXiv.2004.10568
  18. Lin, Tsung-Yi, Maire, Michael, Belongie, Serge, Hays, James, Perona, Pietro, Ramanan, Deva, Dollar, Piotr, & Zitnick, C. Lawrence. "Microsoft coco: Common objects in context." Computer Vision-ECCV 2014: 13th European Conference, pp. 740-755, Zurich, Switzerland, Sep. 2014. DOI: https://doi.org/10.1007/978-3-319-10602-1_48