Vanish into Thin Air: Cross-prompt Universal Adversarial Attacks for SAM2

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zhou, Ziqi, Hu, Yifan, Song, Yufei, Li, Zijing, Hu, Shengshan, Zhang, Leo Yu, Yao, Dezhong, Zheng, Long, Jin, Hai
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909873757749248
author Zhou, Ziqi
Hu, Yifan
Song, Yufei
Li, Zijing
Hu, Shengshan
Zhang, Leo Yu
Yao, Dezhong
Zheng, Long
Jin, Hai
author_facet Zhou, Ziqi
Hu, Yifan
Song, Yufei
Li, Zijing
Hu, Shengshan
Zhang, Leo Yu
Yao, Dezhong
Zheng, Long
Jin, Hai
contents Recent studies reveal the vulnerability of the image segmentation foundation model SAM to adversarial examples. Its successor, SAM2, has attracted significant attention due to its strong generalization capability in video segmentation. However, its robustness remains unexplored, and it is unclear whether existing attacks on SAM can be directly transferred to SAM2. In this paper, we first analyze the performance gap of existing attacks between SAM and SAM2 and highlight two key challenges arising from their architectural differences: directional guidance from the prompt and semantic entanglement across consecutive frames. To address these issues, we propose UAP-SAM2, the first cross-prompt universal adversarial attack against SAM2 driven by dual semantic deviation. For cross-prompt transferability, we begin by designing a target-scanning strategy that divides each frame into k regions, each randomly assigned a prompt, to reduce prompt dependency during optimization. For effectiveness, we design a dual semantic deviation framework that optimizes a UAP by distorting the semantics within the current frame and disrupting the semantic consistency across consecutive frames. Extensive experiments on six datasets across two segmentation tasks demonstrate the effectiveness of the proposed method for SAM2. The comparative results show that UAP-SAM2 significantly outperforms state-of-the-art (SOTA) attacks by a large margin.
format Preprint
id arxiv_https___arxiv_org_abs_2510_24195
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Vanish into Thin Air: Cross-prompt Universal Adversarial Attacks for SAM2
Zhou, Ziqi
Hu, Yifan
Song, Yufei
Li, Zijing
Hu, Shengshan
Zhang, Leo Yu
Yao, Dezhong
Zheng, Long
Jin, Hai
Computer Vision and Pattern Recognition
Recent studies reveal the vulnerability of the image segmentation foundation model SAM to adversarial examples. Its successor, SAM2, has attracted significant attention due to its strong generalization capability in video segmentation. However, its robustness remains unexplored, and it is unclear whether existing attacks on SAM can be directly transferred to SAM2. In this paper, we first analyze the performance gap of existing attacks between SAM and SAM2 and highlight two key challenges arising from their architectural differences: directional guidance from the prompt and semantic entanglement across consecutive frames. To address these issues, we propose UAP-SAM2, the first cross-prompt universal adversarial attack against SAM2 driven by dual semantic deviation. For cross-prompt transferability, we begin by designing a target-scanning strategy that divides each frame into k regions, each randomly assigned a prompt, to reduce prompt dependency during optimization. For effectiveness, we design a dual semantic deviation framework that optimizes a UAP by distorting the semantics within the current frame and disrupting the semantic consistency across consecutive frames. Extensive experiments on six datasets across two segmentation tasks demonstrate the effectiveness of the proposed method for SAM2. The comparative results show that UAP-SAM2 significantly outperforms state-of-the-art (SOTA) attacks by a large margin.
title Vanish into Thin Air: Cross-prompt Universal Adversarial Attacks for SAM2
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2510.24195