arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2104.01732cs.CV
收藏

分割模型的语义隐形对抗攻击

Semantically Stealthy Adversarial Attacks against Segmentation Models

Indiana University Bloomington(印第安纳大学伯明顿分校)

Zhenhua Chen, Chuhua Wang, David J. Crandall

摘要:

已经发现分割模型容易受到有针对性的和非有针对性的对抗攻击。然而,产生的分割输出往往是如此损坏,很容易发现攻击。在本文中,我们提出了语义隐形的对抗攻击,它可以操纵目标标签,同时保留非目标标签。一个挑战是在数据集和模型之间进行语义上有意义的操作。另一个挑战是避免损坏非目标标签。为了解决这些挑战,我们将每个输入图像视为先验知识来生成扰动。我们还设计了一个特殊的正则化器来帮助提取特征。为了评估我们的模型的性能,我们设计了三种基本的攻击类型,即“消失在上下文中”,“嵌入假标签”和“取代目标对象”。“我们的实验表明,我们的隐形对抗模型可以在Cityscapes,Mapillary和BDD 100K上以相对较高的成功率攻击分割模型。我们的框架在数据集和模型中表现出良好的经验概括性。

英文摘要:

Segmentation models have been found to be vulnerable to targeted and non-targeted adversarial attacks. However, the resulting segmentation outputs are often so damaged that it is easy to spot an attack. In this paper, we propose semantically stealthy adversarial attacks which can manipulate targeted labels while preserving non-targeted labels at the same time. One challenge is making semantically meaningful manipulations across datasets and models. Another challenge is avoiding damaging non-targeted labels. To solve these challenges, we consider each input image as prior knowledge to generate perturbations. We also design a special regularizer to help extract features. To evaluate our model's performance, we design three basic attack types, namely `vanishing into the context,' `embedding fake labels,' and `displacing target objects.' Our experiments show that our stealthy adversarial model can attack segmentation models with a relatively high success rate on Cityscapes, Mapillary, and BDD100K. Our framework shows good empirical generalization across datasets and models.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑