Table 1 In-dataset Evaluation on FF++.
From: Localization and detection of deepfake videos based on self-blending method
Method | Training Set | Test Set AUC (%) | ||||
|---|---|---|---|---|---|---|
Real | Fake | FF++DF | FF++F2F | FF++FS | FF++NT | |
Xception36 | \(\checkmark\) | \(\checkmark\) | 96.1 | 92.1 | 93.8 | 85.7 |
F3Net12 | \(\checkmark\) | \(\checkmark\) | 98.0 | 95.2 | 96.0 | 88.4 |
Two-branch19 | \(\checkmark\) | \(\checkmark\) | 98.8 | 96.0 | 97.4 | 88.6 |
LipForensics25 | \(\checkmark\) | \(\checkmark\) | 99.7 | 90.1 | 99.7 | 99.1 |
Face X-Ray15 | \(\checkmark\) | 99.17 | 98.57 | 98.21 | 98.13 | |
Ours | \(\checkmark\) | 99.95 | 98.82 | 99.88 | 99.13 | |
