TAR: Generalized Forensic Framework to Detect Deepfakes using Weakly Supervised Learning

TL;DR

This paper introduces TAR, a transfer learning-based autoencoder framework that effectively detects multiple types of deepfakes with high accuracy and minimal training data, outperforming existing methods in real-world scenarios.

Contribution

The paper presents a novel generalized deepfake detection model using transfer learning and residual autoencoders, capable of identifying various deepfake types simultaneously with limited training samples.

Findings

Achieves higher generalized detection performance than state-of-the-art methods on FaceForensics++

Attains 89.49% zero-shot accuracy on real-world Deepfake-in-the-Wild videos

Demonstrates robustness with minimal training data in practical settings

Abstract

Deepfakes have become a critical social problem, and detecting them is of utmost importance. Also, deepfake generation methods are advancing, and it is becoming harder to detect. While many deepfake detection models can detect different types of deepfakes separately, they perform poorly on generalizing the detection performance over multiple types of deepfake. This motivates us to develop a generalized model to detect different types of deepfakes. Therefore, in this work, we introduce a practical digital forensic tool to detect different types of deepfakes simultaneously and propose Transfer learning-based Autoencoder with Residuals (TAR). The ultimate goal of our work is to develop a unified model to detect various types of deepfake videos with high accuracy, with only a small number of training samples that can work well in real-world settings. We develop an autoencoder-based detection model with Residual blocks and sequentially perform transfer learning to detect different types of deepfakes simultaneously. Our approach achieves a much higher generalized detection performance than the state-of-the-art methods on the FaceForensics++ dataset. In addition, we evaluate our model on 200 real-world Deepfake-in-the-Wild (DW) videos of 50 celebrities available on the Internet and achieve 89.49% zero-shot accuracy, which is significantly higher than the best baseline model (gaining 10.77%), demonstrating and validating the practicability of our approach.