The Journal of the Korea institute of electronic communication sciences (한국전자통신학회논문지)

Korea Institute of Electronic Communication Science (한국전자통신학회)
권호 표지
DOI QR코드

DOI QR Code

Object Tracking Algorithm based on Siamese Network with Local Overlap Confidence

지역 중첩 신뢰도가 적용된 샴 네트워크 기반 객체 추적 알고리즘

  • 임수창 (순천대학교 컴퓨터공학과) ;
  • 김종찬 (순천대학교 컴퓨터공학과)
  • Received : 2023.10.16
  • Accepted : 2023.12.27
  • Published : 2023.12.31
Download PDF
⟨ Previous Next ⟩

Abstract

Object tracking is used to track a goal in a video sequence by using coordinate information provided as annotation in the first frame of the video. In this paper, we propose a tracking algorithm that combines deep features and region inference modules to improve object tracking accuracy. In order to obtain sufficient object information, a convolution neural network was designed with a Siamese network structure. For object region inference, the region proposal network and overlapping confidence module were applied and used for tracking. The performance of the proposed tracking algorithm was evaluated using the Object Tracking Benchmark dataset, and it achieved 69.1% in the Success index and 89.3% in the Precision Metrics.

객체 추적은 영상의 첫 번째 프레임에서 annotation으로 제공되는 좌표 정보를 활용하여 비디오 시퀀스의 목표 추적에 활용된다. 본 논문에서는 객체 추적 정확도 향상을 위해 심층 특징과 영역 추론 모듈을 결합한 추적 알고리즘을 제안한다. 충분한 객체 정보를 획득하기 위해 Convolution Neural Network를 Siamese Network 구조로 네트워크를 설계하였다. 객체의 영역 추론을 위해 지역 제안 네트워크와 중첩 신뢰도 모듈을 적용하여 추적에 활용하였다. 제안한 추적 알고리즘은 Object Tracking Benchmark 데이터셋을 사용하여 성능검증을 수행하였고, Success 지표에서 69.1%, Precision 지표에서 89.3%를 달성하였다.

Keywords

Acknowledgement

본 논문은 2022년도 순천대학교 학술연구비 공모과제로 연구되었음.

References

  1. S. Kim, S. Lim, and D. Kim, "Intelligent Intrusion Detection System Featuring a Virtual Fence, Active Intruder Detection, Classification, Tracking, and Action Recognition," Annals of Nuclear Energy, vol. 112, Feb. 2018, pp. 845-855. https://doi.org/10.1016/j.anucene.2017.11.026
  2. C. Ma, J. B. Huang, X. Yang, and M. H. Yang, "Hierarchical Convolutional Features for Visual Tracking," In Proc. IEEE Int. Conf. on Computer Vision, Santiago, Chile, Dec. 2015, pp. 3074-3082.
  3. P. Li, D. Wang, L. Wang, and H. Lu, "Deep visual tracking: review and experimental comparison," Pattern Recognition, vol. 76, Apr. 2018, pp. 323-338. https://doi.org/10.1016/j.patcog.2017.11.007
  4. S. Lim and D. Kim, "Semantic Segmentation using Convolutional Neural Network with Conditional Random Field," J. of the Korea Institute of Electronic Communication Sciences, vol. 12, no. 3, June 2017, pp. 451-456.
  5. O. Khin and S. Lee, "Feature Extraction and Recognition of Myanmar Characters Based on Deep Learning," J. of the Korea Institute of Electronic Communication Sciences, vol. 17, no. 5, 2023, pp. 977-984.
  6. M. Choi and M. Moon, "Analysis System for Public Interest Report Video of Traffic Law Violation based on Deep Learning Algorithms," J. of the Korea Institute of Electronic Communication Sciences, vol. 18, no. 1, 2023, pp. 63-70.
  7. L. Bertinetto, J. Valmadre, J. F. Henriques, A. Vedaldi, and P.-H. Torr, "Fully-Convolutional Siamese Networks for Object Tracking," In European Conf. on Computer Vision, Amsterdam, Netherlands, Oct 2016, pp. 850-865.
  8. J. Valmadre, L. Bertinetto, J. F. Henriques, A. Vedaldi, and P.H. Torr, "End-to-end representation learning for correlation filter based tracking," In Proc. IEEE conf. on computer vision and pattern recognition, Honolulu, Hawaii, July 2017, pp. 2805-2813.
  9. Q. Guo, W. Feng, C. Zhou, R. Huang, L. Wan, and S. Wang, "Learning dynamic siamese network for visual object tracking," In Proc. IEEE Int. Conf. on Computer Vision, Venice, Italy, Oct 2017, pp. 1763-1771.
  10. B. Li, J. Yan, W. Wu, Z. Zhu, and X. Hu, "High performance visual tracking with siamese region proposal network," In Proc. IEEE Conf. on computer vision and pattern recognition, Salt Lake City, USA, June 2018, pp. 8971-8980.
  11. D. S. Bolme, J. R. Beveridge, B. A. Draper, and Y. M. Lui, "Visual object tracking using adaptive correlation filters," In Proc. IEEE Computer Society Conf. on Computer Vision and Pattern Recognition, San Francisco, CA, USA, June 2010, pp. 2544-2550.
  12. S. Lim and D. Kim, "Object Tracking Using Feature Map from Convolutional Neural Network," J. of Korea Multimedia Society, Feb. 2017, vol. 20, no. 2, pp. 126-133. https://doi.org/10.9717/kmms.2017.20.2.126
  13. M. Danelljan, A. Robinson, F. Shahbaz Khan, and M. Felsberg, "Beyond correlation filters: Learning continuous convolution operators for visual tracking," In European Conf. on Computer Vision, Amsterdam, Netherlands, Oct 2016, pp. pp. 472-488.
  14. M. Danelljan, G. Bhat, F. Shahbaz Khan, and M. Felsberg, "Eco: Efficient convolution operators for tracking," In Proc. IEEE conf. on computer vision and pattern recognition, Honolulu, Hawaii, July 2017, pp. 6638-6646.
  15. S. Lim and D. Kim, "Object Tracking Using Feature Map from Convolutional Neural Network," J. of Korea Multimedia Society, Feb. 2022, vol. 25, no. 2, pp. 247-256.
  16. Z. Zhang, Y. Liu, X. Wang, B. Li, and W. Hu, "Learn to match: Automatic matching network design for visual tracking," In Proc. IEEE/CVF Int. Conf. on Computer Vision, Montreal, BC, Canada, Oct. 2021, pp. 13339-13348.
  17. K. He, X. Zhang, S. Ren, and J. Sun, "Deep residual learning for image recognition," In Proc. IEEE Conf. on Computer Vision and Pattern Recognition, June 2016, Las Vegas, NV, USA, pp. 770-778.
  18. S. Ren, K. He, R. Girshick, and J. Sun, "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks," Advances in Neural Information Processing Systems, Montreal, Canada, Dec. 2015, pp. 91-99.
  19. B. Jiang, R. Luo, J. Mao, T. Xiao, and Y. Jiang, "Acquisition of localization confidence for accurate object detection," In Proc. European Conf. on Computer Vision, Munich, Germany, Sept. 2018, pp. 784-799.
  20. Y. Wu, J. Lim, and M. H. Yang, "Online object tracking: A benchmark," In Proc. IEEE Conf. on Computer Vision and Pattern Recognition, Portland, OR, USA, June 2013, pp. 2411-2418.