Traditional pixel-wise image attack algorithms suffer from poor robustness to defense algorithms, i.e., the attack strength degrades dramatically when defense algorithms are applied. Although Generative Adversarial Networks (GAN) can partially address this problem by synthesizing a more semantically meaningful texture pattern, the main limitation is that existing generators can only generate images of a specific scale. In this paper, we propose a scale-free generation-based attack algorithm that synthesizes semantically meaningful adversarial patterns globally to images with arbitrary scales. Our generative attack approach consistently outperforms the state-of-the-art methods on a wide range of attack settings, i.e. the proposed approach largely degraded the performance of various image classification, object detection, and instance segmentation algorithms under different advanced defense methods.

翻译：传统的像素图像攻击算法对防御算法的强力不强,即攻击强度在应用防御算法时会急剧下降。 虽然创性反versarial 网络(GAN)可以通过合成一个更具有地震意义的质谱模式来部分解决这一问题,但主要限制是现有生成器只能生成特定比例的图像。 在本文中,我们提出了一个无规模的无规模的以一代为基础的攻击算法,将全球范围内具有语义意义的对抗模式与带有任意比例的图像结合起来。 我们的基因攻击方法在广泛的攻击设置上始终优于最先进的方法,即拟议方法在很大程度上降低了不同先进防御方法下各种图像分类、物体探测和实例分割算法的性能。