FRIDAY: Mitigating Unintentional Facial Identity in Deepfake Detectors Guided by Facial Recognizers

TL;DR

FRIDAY is a novel training approach that reduces facial identity influence in deepfake detectors, improving their robustness across different datasets by minimizing identity-related features.

Contribution

We introduce Facial Recognition Identity Attenuation (FRIDAY), a new method that mitigates facial identity bias in deepfake detection models using a face recognizer during training.

Findings

Enhanced cross-domain detection accuracy

Reduced identity bias in deepfake embeddings

Improved robustness on diverse datasets

Abstract

Previous Deepfake detection methods perform well within their training domains, but their effectiveness diminishes significantly with new synthesis techniques. Recent studies have revealed that detection models often create decision boundaries based on facial identity rather than synthetic artifacts, resulting in poor performance on cross-domain datasets. To address this limitation, we propose Facial Recognition Identity Attenuation (FRIDAY), a novel training method that mitigates facial identity influence using a face recognizer. Specifically, we first train a face recognizer using the same backbone as the Deepfake detector. The recognizer is then frozen and employed during the detector's training to reduce facial identity information. This is achieved by feeding input images into both the recognizer and the detector, and minimizing the similarity of their feature embeddings through our Facial Identity Attenuating loss. This process encourages the detector to generate embeddings distinct from the recognizer, effectively reducing the impact of facial identity. Extensive experiments demonstrate that our approach significantly enhances detection performance on both in-domain and cross-domain datasets.