File Download
There are no files associated with this item.
Supplementary
-
Citations:
- Appears in Collections:
Conference Paper: A Unified Multi-Scenario Attacking Network for Visual Object Tracking
Title | A Unified Multi-Scenario Attacking Network for Visual Object Tracking |
---|---|
Authors | |
Keywords | Adversarial Attacks & Robustness Motion & Tracking |
Issue Date | 2021 |
Publisher | AAAI Press. The Journal's web site is located at https://aaai.org/Library/AAAI/aaai-library.php |
Citation | Proceedings of the 35th AAAI Conference on Artificial Intelligence (AAAI), Virtual Confernce, 2-9 February 2021, v. 35 n. 2, p. 1097-1104 How to Cite? |
Abstract | Existing methods of adversarial attacks successfully generate adversarial examples to confuse Deep Neural Networks (DNNs) of image classification and object detection, resulting in wrong predictions. However, these methods are difficult to attack models of video object tracking, because the tracking algorithms could handle sequential information across video frames and the categories of targets tracked are normally unknown in advance. In this paper, we propose a Unified and Effective Network, named UEN, to attack visual object tracking models. There are several appealing characteristics of UEN: (1) UEN could produce various invisible adversarial perturbations according to different attack settings by using only one simple end-to-end network with three ingenious loss function; (2) UEN could generate general visible adversarial patch patterns to attack the advanced trackers in the real-world; (3) Extensive experiments show that UEN is able to attack many state-of-the-art trackers effectively (e.g. SiamRPN-based networks and DiMP) on popular tracking datasets including OTB100, UAV123, and GOT10K, making online real-time attacks possible. The attack results outperform the introduced baseline in terms of attacking ability and attacking efficiency. |
Description | AAAI-21 Technical Tracks 2 / Session: AAAI Technical Track on Computer Vision I |
Persistent Identifier | http://hdl.handle.net/10722/301434 |
ISSN |
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Chen, X | - |
dc.contributor.author | Fu, C | - |
dc.contributor.author | Zheng, F | - |
dc.contributor.author | Zhao, Y | - |
dc.contributor.author | Li, H | - |
dc.contributor.author | Luo, P | - |
dc.contributor.author | QI, G | - |
dc.date.accessioned | 2021-07-27T08:11:00Z | - |
dc.date.available | 2021-07-27T08:11:00Z | - |
dc.date.issued | 2021 | - |
dc.identifier.citation | Proceedings of the 35th AAAI Conference on Artificial Intelligence (AAAI), Virtual Confernce, 2-9 February 2021, v. 35 n. 2, p. 1097-1104 | - |
dc.identifier.issn | 2159-5399 | - |
dc.identifier.uri | http://hdl.handle.net/10722/301434 | - |
dc.description | AAAI-21 Technical Tracks 2 / Session: AAAI Technical Track on Computer Vision I | - |
dc.description.abstract | Existing methods of adversarial attacks successfully generate adversarial examples to confuse Deep Neural Networks (DNNs) of image classification and object detection, resulting in wrong predictions. However, these methods are difficult to attack models of video object tracking, because the tracking algorithms could handle sequential information across video frames and the categories of targets tracked are normally unknown in advance. In this paper, we propose a Unified and Effective Network, named UEN, to attack visual object tracking models. There are several appealing characteristics of UEN: (1) UEN could produce various invisible adversarial perturbations according to different attack settings by using only one simple end-to-end network with three ingenious loss function; (2) UEN could generate general visible adversarial patch patterns to attack the advanced trackers in the real-world; (3) Extensive experiments show that UEN is able to attack many state-of-the-art trackers effectively (e.g. SiamRPN-based networks and DiMP) on popular tracking datasets including OTB100, UAV123, and GOT10K, making online real-time attacks possible. The attack results outperform the introduced baseline in terms of attacking ability and attacking efficiency. | - |
dc.language | eng | - |
dc.publisher | AAAI Press. The Journal's web site is located at https://aaai.org/Library/AAAI/aaai-library.php | - |
dc.relation.ispartof | Proceedings of the AAAI Conference on Artificial Intelligence | - |
dc.subject | Adversarial Attacks & Robustness | - |
dc.subject | Motion & Tracking | - |
dc.title | A Unified Multi-Scenario Attacking Network for Visual Object Tracking | - |
dc.type | Conference_Paper | - |
dc.identifier.email | Luo, P: pluo@hku.hk | - |
dc.identifier.authority | Luo, P=rp02575 | - |
dc.identifier.hkuros | 323760 | - |
dc.identifier.volume | 35 | - |
dc.identifier.issue | 2 | - |
dc.identifier.spage | 1097 | - |
dc.identifier.epage | 1104 | - |
dc.publisher.place | United States | - |