Exploring Strengths and Weaknesses of Super-Resolution Attack in Deepfake Detection

TL;DR

This paper investigates how super-resolution attacks can undermine deepfake detection by hiding artifacts, evaluates their effectiveness across datasets, and proposes training modifications to enhance detector robustness.

Contribution

It explores the impact of super-resolution attacks on deepfake detectors and suggests training adjustments to improve resistance against such adversarial manipulations.

Findings

Super-resolution can hide deepfake artifacts effectively.

Super-resolution fails to conceal traces in fully synthetic images.

Training modifications can improve detector robustness.

Abstract

Image manipulation is rapidly evolving, allowing the creation of credible content that can be used to bend reality. Although the results of deepfake detectors are promising, deepfakes can be made even more complicated to detect through adversarial attacks. They aim to further manipulate the image to camouflage deepfakes' artifacts or to insert signals making the image appear pristine. In this paper, we further explore the potential of super-resolution attacks based on different super-resolution techniques and with different scales that can impact the performance of deepfake detectors with more or less intensity. We also evaluated the impact of the attack on more diverse datasets discovering that the super-resolution process is effective in hiding the artifacts introduced by deepfake generation models but fails in hiding the traces contained in fully synthetic images. Finally, we propose some changes to the detectors' training process to improve their robustness to this kind of attack.