Deepfake Detection on WildDeepfake (test)
0.937AUCMARE
Evaluation results
| Method | Links | ||
|---|---|---|---|
| MAREEvaluation Setup=intra-dataset2026.01 | 0.937 | 0.8722 | |
| M2F2Venue=CVPR25, Evaluation Setup=intra-dataset2026.01 | 0.9314 | 0.8605 | |
| RECCEVenue=CVPR22, Evaluation Setup=intra-dataset2026.01 | 0.9202 | 0.8325 | |
| MultiAttVenue=CVPR21, Evaluation Setup=intra-dataset2026.01 | 0.9071 | 0.8286 | |
| QAD-ETraining set=FF++, Backbone=EfficientNet-B12023.09 | 0.747 | 0.655 | |
| Ours†Training Dataset=FF++, Feature Extracting Backbone=Swin Transformer2026.01 | 0.7313 | — | |
| Ours*Training Dataset=FF++, Feature Extracting Backbone=EfficientNet-B42026.01 | 0.7211 | — | |
| MATTraining set=FF++2023.09 | 0.71 | 0.636 | |
| SBIsTraining Dataset=FF++, Feature Extracting Backbone=Swin Transformer2026.01 | 0.7056 | — | |
| QAD-RTraining set=FF++, Backbone=ResNet-502023.09 | 0.689 | 0.625 | |
| Fang & LinTraining set=FF++2023.09 | 0.677 | 0.631 | |
| Rössler et al.Training set=FF++2023.09 | 0.664 | 0.611 | |
| SBIsTraining set=FF++2023.09 | 0.655 | 0.592 | |
| F3NetTraining set=FF++2023.09 | 0.645 | 0.589 | |
| RECCETraining Dataset=FF++2026.01 | 0.6431 | — | |
| EfficientNet-B4Training Dataset=FF++2026.01 | 0.6383 | — | |
| XceptionTraining Dataset=FF++2026.01 | 0.6272 | — | |
| Multi-AttTraining Dataset=FF++2026.01 | 0.5974 | — | |
| MesoNetTraining set=FF++2023.09 | 0.558 | 0.544 |