A lightweight XGBoost classifier trained on statistical moments of pairwise ArcFace biometric similarities can distinguish real talking-head videos from face-swap and lip-sync deepfakes with around 95% accuracy on matched and combined datasets.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Detecting Deepfake Talking Heads from Facial Biometric Anomalies
A lightweight XGBoost classifier trained on statistical moments of pairwise ArcFace biometric similarities can distinguish real talking-head videos from face-swap and lip-sync deepfakes with around 95% accuracy on matched and combined datasets.