Detecting Deepfake Talking Heads from Facial Biometric Anomalies
This addresses the issue of fraud, scams, and political disinformation caused by deepfake impersonations, representing an incremental improvement in detection methods.
The paper tackles the problem of detecting deepfake talking head videos by proposing a forensic machine learning technique that identifies unnatural patterns in facial biometrics, achieving evaluation across a large dataset of deepfake techniques and assessing reliability to video laundering and generalization to unseen generators.
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.