Journal of Computer Research and Development
|
2024, 61(2): 460-469
• Artificial Intelligence and Pattern Recognition •
Mixing Tokens from Target and Search Regions for Visual Object Tracking
Full
Wanli Xue
1, Zhibin Zhang
1, Shenglei Pei
2, *
, Kaihua Zhang
3, Shengyong Chen
1
Affiliations
- 1School of Computer Science and Engineering, Tianjin University of Technology, Tianjin 300384
- 2College of Physics and Electronic Information Engineering, Qinghai Minzu University, Xining 810007
- 3School of Computer Science, Nanjing University of Information Science & Technology, Nanjing 130012
born in 1986. PhD, associate professor, master supervisor. Member of CCF and CSIG. His main research interests include visual tracking, sign language recognition, and image stitching
born in 1996. PhD candidate. His main research interests include visual object tracking and deep learning
born in 1980. PhD, professor, master supervisor. His main research interests include machine learning, data mining, and intelligent decision system
born in 1983. PhD, professor. His main research interests include video object segmentation and visual object tracking
born in 1973. PhD, professor. His main research interests include computer vision and machine learning
About Author:
born in 1986. PhD, associate professor, master supervisor. Member of CCF and CSIG. His main research interests include visual tracking, sign language recognition, and image stitching
born in 1996. PhD candidate. His main research interests include visual object tracking and deep learning
born in 1980. PhD, professor, master supervisor. His main research interests include machine learning, data mining, and intelligent decision system
born in 1983. PhD, professor. His main research interests include video object segmentation and visual object tracking
born in 1973. PhD, professor. His main research interests include computer vision and machine learning
Published: 2024-02-02
doi: 10.7544/issn1000-1239.202220698
Outline
There are three problems about feature extraction and fusion in the current mainstream tracking framework based on Transformer: 1. The two modules of feature extraction and fusion are used separately, which is easy to produce sub-optimal model training results. 2. Computational complexity of $ O\left({N}^{2}\right) $ using self-attention reduces tracking efficiency. 3. The target template selection strategy is simple and is difficult to adapt to the drastic changes in the target appearance during the tracking process. We propose a novel Transformer tracking framework using fast Fourier transform mixing target tokens and search region tokens. For problem 1, an efficient end-to-end approach is proposed to extract and fuse features for unified learning to obtain optimal model; For problem 2, the fast Fourier transform is used to achieve complete information interaction between the target tokens and search region tokens. The computational complexity of this operation is $ O\left(N\mathrm{l}\mathrm{o}\mathrm{g}\left(N\right)\right) $, which greatly improves the tracing efficiency. For problem 3, a template memory storage mechanism based on quality assessment is proposed, which can quickly adapt to the drastic changes in target appearance. Compared with the current state-of-the-art algorithms on three datasets LaSOT, OTB100 and UAV123, our tracker achieves better performance in both efficiency and accuracy.
Transformer
/
fast Fourier transform
/
feature extraction
/
feature fusion
/
object tracking
薛万利, 张智彬, 裴生雷, 张开华, 陈胜勇.
混合目标与搜索区域令牌的视觉目标跟踪.
计算机研究与发展,
2024
, 61
(2)
: 460
-469
.
DOI: 10.7544/issn1000-1239.202220698
Wanli Xue, Zhibin Zhang, Shenglei Pei, Kaihua Zhang, Shengyong Chen.
Mixing Tokens from Target and Search Regions for Visual Object Tracking[J].
Journal of Computer Research and Development,
2024
, 61
(2)
: 460
-469
.
DOI: 10.7544/issn1000-1239.202220698
Year 2024 volume 61 Issue 2
73
35
Cite this Article
BibTeX
Article Info
doi: 10.7544/issn1000-1239.202220698
- Receive Date:2022-08-08
- Online Date:2026-11-17
- Published:2024-02-02