Lightweight Transformer Tracker: Compact and Effect Neural Network for Object Tracking with Long-Short Range Attention
Pages 384 - 390
Abstract
Recent years, attention mechanism has been widely used in computer vision, such as object detection, tracking and recognition. Many studies reveal that attention has better performance than those CNN-based or RNN-based networks. However, the cost of better performance, such as complicated structure and complex algorithm, cannot be neglected. In this paper, a lightweight transformer tracker named LTT is proposed. Different from Transformer Tracking network (TransT), three aspects of lightweight operation are adopted: firstly, YOLO-nano Darknet is utilized as the feature extraction network; secondly, the size of template image is scaled to 1/4 of the original and removing the self-attention layer; finally, the combination layer of convolution and cross-attention (long-short range attention) is adopted for the sake of feature fusion. Experiments show that our tracker runs at roughly 70 fps on GPU while there is no significant performance loss compared with other networks like SiamFc or SiamRPN++. Moreover, the model size of our tracker is only 4.58M.
References
[1]
CHEN X, YAN B, ZHU J, Transformer Tracking[M/OL]. arXiv, 2021[2022-06-03]. http://arxiv.org/abs/2103.15436.
[2]
VASWANI A, SHAZEER N, PARMAR N, Attention Is All You Need[J/OL]. CoRR, 2017, abs/1706.03762. http://arxiv.org/abs/1706.03762.
[3]
WU Z, LIU Z, LIN J, Lite Transformer with Long-Short Range Attention[J/OL]. CoRR, 2020, abs/2004.11886. https://arxiv.org/abs/2004.11886.
[4]
GE Z, LIU S, WANG F, YOLOX: Exceeding YOLO Series in 2021[J/OL]. CoRR, 2021, abs/2107.08430. https://arxiv.org/abs/2107.08430.
[5]
HUANG L, ZHAO X, HUANG K. GOT-10k: A Large High-Diversity Benchmark for Generic Object Tracking in the Wild[J/OL]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 43(5): 1562-1577. https://doi.org/10.1109/TPAMI.2019.2957464.
[6]
Object Tracking Benchmark | IEEE Journals & Magazine | IEEE Xplore[EB/OL]. [2022-06-03]. https://ieeexplore.ieee.org/document/7001050.
[7]
A Benchmark and Simulator for UAV Tracking | SpringerLink[EB/OL]. [2022-06-02]. https://link.springer.com/chapter/10.1007/978-3-319-46448-0_27.
[8]
XING D, EVANGELIOU N, TSOUKALAS A, Siamese Transformer Pyramid Networks for Real-Time UAV Tracking: arXiv:2110.08822[R/OL]. arXiv, 2021[2022-06-02]. http://arxiv.org/abs/2110.08822.
[9]
WANG N, ZHOU W, WANG J, Transformer Meets Tracker: Exploiting Temporal Context for Robust Visual Tracking[J/OL]. CoRR, 2021, abs/2103.11681. https://arxiv.org/abs/2103.11681.
[10]
CARION N, MASSA F, SYNNAEVE G, End-to-End Object Detection with Transformers[M/OL]. arXiv, 2020[2022-08-17]. http://arxiv.org/abs/2005.12872.
[11]
LaSOT: A High-Quality Benchmark for Large-Scale Single Object Tracking | IEEE Conference Publication | IEEE Xplore[EB/OL]. [2022-06-03]. https://ieeexplore.ieee.org/document/8954084/.
[12]
[1812.11703] SiamRPN++: Evolution of Siamese Visual Tracking with Very Deep Networks[EB/OL]. [2022-06-03]. https://arxiv.org/abs/1812.11703.
[13]
BERTINETTO L, VALMADRE J, HENRIQUES J F, Fully-Convolutional Siamese Networks for Object Tracking[M/OL]//HUA G, JÉGOU H. Computer Vision – ECCV 2016 Workshops: 9914. Cham: Springer International Publishing, 2016: 850-865[2022-08-17]. http://link.springer.com/10.1007/978-3-319-48881-3_56.
[14]
Distractor-Aware Siamese Networks for Visual Object Tracking | SpringerLink[EB/OL]. [2022-06-03]. https://link.springer.com/chapter/10.1007/978-3-030-01240-3_7.
[15]
CHEN Q, WANG Y, YANG T, You Only Look One-level Feature[J/OL]. CoRR, 2021, abs/2103.09460. https://arxiv.org/abs/2103.09460.
[16]
DANELLJAN M, VAN GOOL L, TIMOFTE R. Probabilistic Regression for Visual Tracking[M/OL]. arXiv, 2020[2022-08-17]. http://arxiv.org/abs/2003.12565.
[17]
BHAT G, DANELLJAN M, VAN GOOL L, Learning Discriminative Model Prediction for Tracking[C/OL]//2019 IEEE/CVF International Conference on Computer Vision (ICCV). Seoul, Korea (South): IEEE, 2019: 6181-6190[2022-08-17]. https://ieeexplore.ieee.org/document/9010649/.
[18]
DANELLJAN M, BHAT G, KHAN F S, ATOM: Accurate Tracking by Overlap Maximization[M/OL]. arXiv, 2019[2022-08-17]. http://arxiv.org/abs/1811.07628.
[19]
DANELLJAN M, BHAT G, KHAN F S, ECO: Efficient Convolution Operators for Tracking[M/OL]. arXiv, 2017[2022-08-17]. http://arxiv.org/abs/1611.09224.
[20]
NAM H, HAN B. Learning Multi-Domain Convolutional Neural Networks for Visual Tracking[M/OL]. arXiv, 2016[2022-08-17]. http://arxiv.org/abs/1510.07945.
[21]
CHEN Z, ZHONG B, LI G, Siamese Box Adaptive Network for Visual Tracking: arXiv:2003.06761[R/OL]. arXiv, 2020[2022-06-02]. http://arxiv.org/abs/2003.06761.
[22]
GUO D, WANG J, CUI Y, SiamCAR: Siamese Fully Convolutional Classification and Regression for Visual Tracking: arXiv:1911.07241[R/OL]. arXiv, 2019[2022-06-02]. http://arxiv.org/abs/1911.07241.
[23]
LI B, YAN J, WU W, High Performance Visual Tracking with Siamese Region Proposal Network[C/OL]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City, UT: IEEE, 2018: 8971-8980[2022-06-02]. https://ieeexplore.ieee.org/document/8579033/.
[24]
WANG Q, ZHANG L, BERTINETTO L, Fast Online Object Tracking and Segmentation: A Unifying Approach: arXiv:1812.05050[R/OL]. arXiv, 2019[2022-06-02]. http://arxiv.org/abs/1812.05050.
Index Terms
- Lightweight Transformer Tracker: Compact and Effect Neural Network for Object Tracking with Long-Short Range Attention
Recommendations
Robust object tracking via multi-cue fusion
A long-term object tracking method based on calibrated binocular cameras by fusing information of the two channels and binocular geometry constraints is proposed.The stereo filter which is built based on the epipolar geometry of the binocular cameras is ...
Twofold Siamese Network with Attentional Feature Fusion for Object Tracking
ICCAI '22: Proceedings of the 8th International Conference on Computing and Artificial IntelligenceObject tracking is still a critical and challenging problem in computer vision. More and more researchers pay attention to applying deep learning to obtain the powerful feature for robust tracking. Nowadays, feature fusion is an essential part of ...
Object Tracking Algorithm for Siamese Network Combined with Channel Attention Mechanism
ICIAI '22: Proceedings of the 2022 6th International Conference on Innovation in Artificial IntelligenceAs an important branch in the field of computer vision, object tracking has been widely used in many fields such as intelligent video surveillance, human-computer interaction and autonomous driving. Although object tracking has imposing development in ...
Comments
Information & Contributors
Information
Published In
September 2022
1221 pages
ISBN:9781450396899
DOI:10.1145/3573942
Copyright © 2022 ACM.
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from [email protected]
Publisher
Association for Computing Machinery
New York, NY, United States
Publication History
Published: 16 May 2023
Check for updates
Author Tags
Qualifiers
- Research-article
- Research
- Refereed limited
Funding Sources
- National Science Foundation of China
- Shaanxi Industrial Development Key Project
Conference
AIPR 2022
AIPR 2022: 2022 5th International Conference on Artificial Intelligence and Pattern Recognition
September 23 - 25, 2022
Xiamen, China
Contributors
Other Metrics
Bibliometrics & Citations
Bibliometrics
Article Metrics
- 0Total Citations
- 43Total Downloads
- Downloads (Last 12 months)19
- Downloads (Last 6 weeks)2
Reflects downloads up to 01 Mar 2025
Other Metrics
Citations
View Options
Login options
Check if you have access through your login credentials or your institution to get full access on this article.
Sign inFull Access
View options
View or Download as a PDF file.
PDFeReader
View online with eReader.
eReaderHTML Format
View this article in HTML Format.
HTML Format