FreqBlender: Enhancing DeepFake Detection by Blending Frequency Knowledge

TL;DR

FreqBlender introduces a novel frequency-based face blending technique that improves DeepFake detection by better simulating forgery traces in the frequency domain, enhancing model generalization.

Contribution

The paper proposes a Frequency Parsing Network and a training strategy to generate pseudo-fake faces by blending frequency components, addressing limitations of spatial domain methods.

Findings

Enhanced DeepFake detection accuracy

Effective pseudo-fake face generation in frequency domain

Potential as a plug-and-play strategy for other methods

Abstract

Generating synthetic fake faces, known as pseudo-fake faces, is an effective way to improve the generalization of DeepFake detection. Existing methods typically generate these faces by blending real or fake faces in spatial domain. While these methods have shown promise, they overlook the simulation of frequency distribution in pseudo-fake faces, limiting the learning of generic forgery traces in-depth. To address this, this paper introduces {\em FreqBlender}, a new method that can generate pseudo-fake faces by blending frequency knowledge. Concretely, we investigate the major frequency components and propose a Frequency Parsing Network to adaptively partition frequency components related to forgery traces. Then we blend this frequency knowledge from fake faces into real faces to generate pseudo-fake faces. Since there is no ground truth for frequency components, we describe a dedicated training strategy by leveraging the inner correlations among different frequency knowledge to instruct the learning process. Experimental results demonstrate the effectiveness of our method in enhancing DeepFake detection, making it a potential plug-and-play strategy for other methods.