Conditional generative adversarial siamese networks for object tracking

被引：0
|
作者：
Song J.-H. [1 ]
Zhang J. [1 ]
Liu Y.-J. [1 ]
Yu Y. [1 ]
机构：
[1] School of Automation and Electrical Engineering, Shenyang Ligong University, Shenyang
来源：
Kongzhi yu Juece/Control and Decision | 2021年 / 36卷 / 05期
关键词：
Conditional generation adversarial network; Fully convolutional siamese network; Motion blur; Target tracking;
D O I：
10.13195/j.kzyjc.2019.1215
中图分类号：
学科分类号：
摘要：
In order to solve the problem that the model drifts due to the motion blur and low resolution when the tracked target moves fast and violently, which leads to the poor tracking effect of the tracker or even the tracking failure, this paper improves the fully convolutional siamese networks for object tracking (SiamFC), and proposes a target tracking algorithm based on conditional generative adversarial siamese networks for object tracking (CGANSiamFC). Firstly, the conditional generative adversarial network module is embedded on the basis of the SiamFC framework to deblur the input low-resolution blurred video frames. Then, the fully convolutional siamese networks perform feature extraction on the reconstructed video frame, which improve the model's characterization ability. Finally, this paper uses separate training and online combination methods to train and test the improved tracking algorithm, and uses the visual tracking benchmark data set OTB100 to evaluate the performance of the improved tracking algorithm. At the same time, in order to fully verify the effectiveness of the proposed algorithm, this paper uses the traditional Lucy-Richardson deblurring algorithm to improve the SiamFC, which is compared with the CGANSiamFC. The experimental results show that the comprehensive accuracy and success rate indicators of the CGANSiamFC proposed are 9% and 8% higher than that of the original SiamFC, respectively, and it has good tracking effects on motion blur and low-resolution moving targets. Copyright ©2021 Control and Decision.
引用
收藏
页码：1110 / 1118
页数：8
相关论文
共 20 条
  • [1] Ross D A, Lim J, Lin R S, Et al., Incremental learning for robust visual tracking, International Journal of Computer Vision, 77, 1, pp. 125-141, (2008)
  • [2] Comaniciu D, Ramesh V, Meer P., Real-time tracking of non-rigid objects using mean shift, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Hilton Head, 2, pp. 142-149, (2000)
  • [3] Watanabe T, Ito S, Yokoi K., Co-occurrence histograms of oriented gradients for human detection, IPSJ Transactions on Computer Vision and Applications, 2, pp. 39-47, (2010)
  • [4] Bolme D S, Beveridge J R, Draper B A, Et al., Visual object tracking using adaptive correlation filters, 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pp. 2544-2550, (2010)
  • [5] Danelljan M, Shahbaz Khan F, Felsberg M, Et al., Adaptive color attributes for real-time visual tracking, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 1090-1097, (2014)
  • [6] Henriques J F, Caseiro R, Martins P, Et al., High-speed tracking with kernelized correlation filters, IEEE Transactions on Pattern Analysis and Machine Intelligence, 37, 3, pp. 583-596, (2015)
  • [7] Ma C, Huang J B, Yang X K, Et al., Hierarchical convolutional features for visual tracking, Proceedings of the IEEE International Conference on Computer Vision, pp. 3074-3082, (2015)
  • [8] Wang L J, Ouyang W L, Wang X G, Et al., Visual tracking with fully convolutional networks, Proceedings of the IEEE International Conference on Computer Vision, pp. 3119-3127, (2015)
  • [9] Nam H, Han B., Learning multi-domain convolutional neural networks for visual tracking, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 4293-4302, (2016)
  • [10] Bertinetto L, Valmadre J, Henriques J F, Et al., Fully-convolutional siamese networks for object tracking, European Conference on Computer Vision, pp. 850-865, (2016)
12