Explicit Correlation Learning for Generalizable Cross-Modal Deepfake Detection

TL;DR

This paper introduces a novel correlation distillation approach to improve the generalizability of cross-modal deepfake detection, supported by a new comprehensive dataset and experimental validation.

Contribution

It proposes a correlation distillation task to explicitly model cross-modal content correlation, enhancing detection across diverse deepfake generation methods.

Findings

Outperforms existing methods on CMDFD and FakeAVCeleb datasets.

Demonstrates improved generalizability across multiple deepfake types.

Provides a new dataset for cross-modal deepfake detection evaluation.

Abstract

With the rising prevalence of deepfakes, there is a growing interest in developing generalizable detection methods for various types of deepfakes. While effective in their specific modalities, traditional detection methods fall short in addressing the generalizability of detection across diverse cross-modal deepfakes. This paper aims to explicitly learn potential cross-modal correlation to enhance deepfake detection towards various generation scenarios. Our approach introduces a correlation distillation task, which models the inherent cross-modal correlation based on content information. This strategy helps to prevent the model from overfitting merely to audio-visual synchronization. Additionally, we present the Cross-Modal Deepfake Dataset (CMDFD), a comprehensive dataset with four generation methods to evaluate the detection of diverse cross-modal deepfakes. The experimental results on CMDFD and FakeAVCeleb datasets demonstrate the superior generalizability of our method over existing state-of-the-art methods. Our code and data can be found at \url{https://github.com/ljj898/CMDFD-Dataset-and-Deepfake-Detection}.