When Generative Replay Meets Evolving Deepfakes: Domain-Aware Relative Weighting for Incremental Face Forgery Detection

TL;DR

This paper introduces a domain-aware weighting strategy for incremental face forgery detection that effectively leverages generative replay, addressing domain overlap issues and improving model performance.

Contribution

It proposes a novel Domain-Aware Relative Weighting (DARW) method that dynamically balances supervision of safe samples and management of risky samples in generative replay.

Findings

DARW improves incremental forgery detection accuracy across various settings.

The method effectively mitigates domain overlap issues in generative replay.

Experimental results show consistent performance gains over existing approaches.

Abstract

The rapid advancement of face generation techniques has led to a growing variety of forgery methods. Incremental forgery detection aims to gradually update existing models with new forgery data, yet current sample replay-based methods are limited by low diversity and privacy concerns. Generative replay offers a potential solution by synthesizing past data, but its feasibility for forgery detection remains unclear. In this work, we systematically investigate generative replay and identify two scenarios: when the replay generator closely resembles the new forgery model, generated real samples blur the domain boundary, creating domain-risky samples; when the replay generator differs significantly, generated samples can be safely supervised, forming domain-safe samples. To exploit generative replay effectively, we propose a novel Domain-Aware Relative Weighting (DARW) strategy. DARW directly supervises domain-safe samples while applying a Relative Separation Loss to balance supervision and potential confusion for domain-risky samples. A Domain Confusion Score dynamically adjusts this tradeoff according to sample reliability. Extensive experiments demonstrate that DARW consistently improves incremental learning performance for forgery detection under different generative replay settings and alleviates the adverse impact of domain overlap.