CASIA OpenIR  > 多模态人工智能系统全国重点实验室  > 视频内容安全
A Simple and Strong Baseline for Universal Targeted Attacks on Siamese Visual Tracking
Li, Zhenbang1,2; Shi, Yaya3; Gao, Jin1,2; Wang, Shaoru1,2; Li, Bing1,2; Liang, Pengpeng4; Hu, Weiming2,5
Source PublicationIEEE Transactions on Circuits and Systems for Video Technology
2022
Volume32Issue:6Pages:3880-3894
Abstract

Siamese trackers are shown to be vulnerable to adversarial attacks recently. However, the existing attack methods craft the perturbations for each video independently, which comes at a non-negligible computational cost. In this paper, we show the existence of universal perturbations that can enable the targeted attack, e.g., forcing a tracker to follow the ground-Truth trajectory with specified offsets, to be video-Agnostic and free from inference in a network. Specifically, we attack a tracker by adding a universal translucent perturbation to the template image and adding a fake target, i.e., a small universal adversarial patch, into the search images adhering to the predefined trajectory, so that the tracker outputs the location and size of the fake target instead of the real target. Our approach allows perturbing a novel video to come at no additional cost except the mere addition operations-and not require gradient optimization or network inference. Experimental results on several datasets demonstrate that our approach can effectively fool the Siamese trackers in a targeted attack manner. We show that the proposed perturbations are not only universal across videos, but also generalize well across different trackers. Such perturbations are therefore doubly universal, both with respect to the data and the network architectures. Our code is available at https://github.com/lizhenbang56/SiamAttack.

Indexed BySCI
Language英语
Sub direction classification目标检测、跟踪与识别
planning direction of the national heavy laboratory实体人工智能系统感认知
Paper associated data
Document Type期刊论文
Identifierhttp://ir.ia.ac.cn/handle/173211/57499
Collection多模态人工智能系统全国重点实验室_视频内容安全
Corresponding AuthorGao, Jin
Affiliation1.Institute of Automation, Chinese Academy of Sciences, National Laboratory of Pattern Recognition, Beijing; 100190, China
2.University of Chinese Academy of Sciences, School of Artificial Intelligence, Beijing; 100190, China
3.University of Science and Technology of China, Hefei, School of Information Science and Technology, Anhui; 230026, China
4.Zhengzhou University, Zhengzhou, School of Information Engineering, Henan; 450001, China
5.CAS Center for Excellence in Brain Science and Intelligence Technology, Institute of Automation, Chinese Academy of Sciences, National Laboratory of Pattern Recognition, Beijing; 100190, China
First Author AffilicationInstitute of Automation, Chinese Academy of Sciences
Corresponding Author AffilicationInstitute of Automation, Chinese Academy of Sciences
Recommended Citation
GB/T 7714
Li, Zhenbang,Shi, Yaya,Gao, Jin,et al. A Simple and Strong Baseline for Universal Targeted Attacks on Siamese Visual Tracking[J]. IEEE Transactions on Circuits and Systems for Video Technology,2022,32(6):3880-3894.
APA Li, Zhenbang.,Shi, Yaya.,Gao, Jin.,Wang, Shaoru.,Li, Bing.,...&Hu, Weiming.(2022).A Simple and Strong Baseline for Universal Targeted Attacks on Siamese Visual Tracking.IEEE Transactions on Circuits and Systems for Video Technology,32(6),3880-3894.
MLA Li, Zhenbang,et al."A Simple and Strong Baseline for Universal Targeted Attacks on Siamese Visual Tracking".IEEE Transactions on Circuits and Systems for Video Technology 32.6(2022):3880-3894.
Files in This Item: Download All
File Name/Size DocType Version Access License
TCSVT2022.pdf(4397KB)期刊论文作者接受稿开放获取CC BY-NC-SAView Download
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[Li, Zhenbang]'s Articles
[Shi, Yaya]'s Articles
[Gao, Jin]'s Articles
Baidu academic
Similar articles in Baidu academic
[Li, Zhenbang]'s Articles
[Shi, Yaya]'s Articles
[Gao, Jin]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[Li, Zhenbang]'s Articles
[Shi, Yaya]'s Articles
[Gao, Jin]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: TCSVT2022.pdf
Format: Adobe PDF
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.