Adversarial Images through Stega Glasses

TL;DR

This paper investigates how steganography techniques can be used to create imperceptible adversarial images and examines their impact on computer vision classifiers and steganalysis tools.

Contribution

It introduces methods to generate undetectable adversarial images using steganography and analyzes their effectiveness against classifiers and steganalyzers.

Findings

Steganography enhances adversarial attack stealth.

Steganalysis is less effective against steganography-based adversarial images.

Steganography tools are more beneficial for attackers than defenders.

Abstract

This paper explores the connection between steganography and adversarial images. On the one hand, ste-ganalysis helps in detecting adversarial perturbations. On the other hand, steganography helps in forging adversarial perturbations that are not only invisible to the human eye but also statistically undetectable. This work explains how to use these information hiding tools for attacking or defending computer vision image classification. We play this cat and mouse game with state-of-art classifiers, steganalyzers, and steganographic embedding schemes. It turns out that steganography helps more the attacker than the defender.