Continual Face Forgery Detection via Historical Distribution Preserving

TL;DR

This paper introduces a continual face forgery detection framework that preserves historical data distributions using adversarial perturbations and knowledge distillation, enabling efficient learning of new attacks without forgetting previous ones.

Contribution

It proposes the Historical Distribution Preserving (HDP) framework for continual face forgery detection, addressing the challenge of learning from new attacks while retaining past knowledge.

Findings

Outperforms state-of-the-art methods on new benchmarks

Effectively preserves historical face distributions during continual learning

Demonstrates robustness against evolving forgery techniques

Abstract

Face forgery techniques have advanced rapidly and pose serious security threats. Existing face forgery detection methods try to learn generalizable features, but they still fall short of practical application. Additionally, finetuning these methods on historical training data is resource-intensive in terms of time and storage. In this paper, we focus on a novel and challenging problem: Continual Face Forgery Detection (CFFD), which aims to efficiently learn from new forgery attacks without forgetting previous ones. Specifically, we propose a Historical Distribution Preserving (HDP) framework that reserves and preserves the distributions of historical faces. To achieve this, we use universal adversarial perturbation (UAP) to simulate historical forgery distribution, and knowledge distillation to maintain the distribution variation of real faces across different models. We also construct a new benchmark for CFFD with three evaluation protocols. Our extensive experiments on the benchmarks show that our method outperforms the state-of-the-art competitors.