Enhancing Abnormality Identification: Robust Out-of-Distribution Strategies for Deepfake Detection

TL;DR

This paper introduces two innovative Out-Of-Distribution detection methods to improve deepfake detection, addressing the challenge of generalizing to unseen generative models in open-set scenarios.

Contribution

It presents two novel OOD detection approaches, one based on image reconstruction and another using attention mechanisms, enhancing deepfake detection robustness.

Findings

Outperforms existing state-of-the-art techniques

Achieves top rankings on benchmark datasets

Demonstrates robustness in real-world scenarios

Abstract

Detecting deepfakes has become a critical challenge in Computer Vision and Artificial Intelligence. Despite significant progress in detection techniques, generalizing them to open-set scenarios continues to be a persistent difficulty. Neural networks are often trained on the closed-world assumption, but with new generative models constantly evolving, it is inevitable to encounter data generated by models that are not part of the training distribution. To address these challenges, in this paper, we propose two novel Out-Of-Distribution (OOD) detection approaches. The first approach is trained to reconstruct the input image, while the second incorporates an attention mechanism for detecting OODs. Our experiments validate the effectiveness of the proposed approaches compared to existing state-of-the-art techniques. Our method achieves promising results in deepfake detection and ranks among the top-performing configurations on the benchmark, demonstrating their potential for robust, adaptable solutions in dynamic, real-world applications.