Think Twice before Adaptation: Improving Adaptability of DeepFake Detection via Online Test-Time Adaptation

TL;DR

This paper introduces T2A, a novel online test-time adaptation method that improves DeepFake detectors' robustness against distribution shifts and postprocessing artifacts without needing source data or labels.

Contribution

The paper proposes T2A, an online adaptation approach using negative learning, uncertainty prioritization, and gradient masking to enhance DeepFake detection in real-world scenarios.

Findings

Achieves state-of-the-art results on DeepFake detection benchmarks.

Significantly improves detector resilience against postprocessing manipulations.

Demonstrates theoretical advantages of negative learning over entropy minimization.

Abstract

Deepfake (DF) detectors face significant challenges when deployed in real-world environments, particularly when encountering test samples deviated from training data through either postprocessing manipulations or distribution shifts. We demonstrate postprocessing techniques can completely obscure generation artifacts presented in DF samples, leading to performance degradation of DF detectors. To address these challenges, we propose Think Twice before Adaptation (\texttt{T$^2$A}), a novel online test-time adaptation method that enhances the adaptability of detectors during inference without requiring access to source training data or labels. Our key idea is to enable the model to explore alternative options through an Uncertainty-aware Negative Learning objective rather than solely relying on its initial predictions as commonly seen in entropy minimization (EM)-based approaches. We also introduce an Uncertain Sample Prioritization strategy and Gradients Masking technique to improve the adaptation by focusing on important samples and model parameters. Our theoretical analysis demonstrates that the proposed negative learning objective exhibits complementary behavior to EM, facilitating better adaptation capability. Empirically, our method achieves state-of-the-art results compared to existing test-time adaptation (TTA) approaches and significantly enhances the resilience and generalization of DF detectors during inference. Code is available \href{https://github.com/HongHanh2104/T2A-Think-Twice-Before-Adaptation}{here}.