English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Person re-identification (re-ID) is the task of matching person images acrosscamera views, which plays an important role in surveillance and securityapplications. Inspired by great progress of deep learning, deep re-ID modelsbegan to be popular and gained state-of-the-art performance. However, recentworks found that deep neural networks (DNNs) are vulnerable to adversarialexamples, posing potential threats to DNNs based applications. This phenomenonthrows a serious question about whether deep re-ID based systems are vulnerableto adversarial attacks. In this paper, we take the first attempt to implementrobust physical-world attacks against deep re-ID. We propose a novel attackalgorithm, called advPattern, for generating adversarial patterns on clothes,which learns the variations of image pairs across cameras to pull closer theimage features from the same camera, while pushing features from differentcameras farther. By wearing our crafted "invisible cloak", an adversary canevade person search, or impersonate a target person to fool deep re-ID modelsin physical world. We evaluate the effectiveness of our transformable patternson adversaries'clothes with Market1501 and our established PRCS dataset. Theexperimental results show that the rank-1 accuracy of re-ID models for matchingthe adversary decreases from 87.9% to 27.1% under Evading Attack. Furthermore,the adversary can impersonate a target person with 47.1% rank-1 accuracy and67.9% mAP under Impersonation Attack. The results demonstrate that deep re-IDsystems are vulnerable to our physical attacks.

advPattern: Physical-World Attacks on Deep Person Re-Identification via Adversarially Transformable Patterns