Nov 6, 2019 · In this work, we drop this assumption by pursuing an approach that exploits human color perception, and more specifically, minimizing perturbation size.

In this work, we propose to create adversarial exam- ples by perturbing images with respect to perceptual color. (PerC) distance. Using PerC distance makes it ...

We propose to strategically relax tight Lp-norm constraints while still maintaining imperceptibility by using perceptual color distance (CIEDE2000).

Perceptual color (PerC) distance allows larger RGB perturbations (cf. L2 and L∞ norm in middle row), while also contributing to imperceptibility (bottom row). ( ...

Experimental evaluation shows PerC approaches outperform conventional Lp approaches in terms of robustness and transferability, and also demonstrates that ...

In this work, we drop this assumption by pursuing an approach that exploits human color perception, and more specifically, minimizing perturbation size.

Nov 6, 2019 · In this work, we investigate the contribution of human color perception to perturbations that are not noticeable. Our basic insight is that ...

The success of image perturbations that are designed to fool image classifier is assessed in terms of both adversarial effect and visual imperceptibility.

Jun 19, 2024 · Towards Large Yet Imperceptible Adversarial Image Perturbations With Perceptual Color Distance. Zhengyu Zhao, Zhuoran Liu, M. Larson. 2019 ...

Introduces a novel method to incorporate perceptual color distances into adversarial attacks. Adversarial examples constructed with this method have large ...