Detecting Deepfake Talking Heads from Facial Biometric Anomalies

TL;DR

This paper introduces a new machine learning method that detects deepfake talking head videos by identifying unnatural facial biometric patterns, effectively distinguishing genuine videos from sophisticated deepfake impersonations.

Contribution

The study presents a novel forensic approach leveraging facial biometric anomalies, with extensive evaluation across diverse deepfake techniques and robustness tests.

Findings

High accuracy in detecting deepfake videos

Effective against unseen deepfake generators

Robust to video laundering techniques

Abstract

The combination of highly realistic voice cloning, along with visually compelling avatar, face-swap, or lip-sync deepfake video generation, makes it relatively easy to create a video of anyone saying anything. Today, such deepfake impersonations are often used to power frauds, scams, and political disinformation. We propose a novel forensic machine learning technique for the detection of deepfake video impersonations that leverages unnatural patterns in facial biometrics. We evaluate this technique across a large dataset of deepfake techniques and impersonations, as well as assess its reliability to video laundering and its generalization to previously unseen video deepfake generators.