Deepfake Detection on FaceShifter (FSH) (test)
99.6AUCRealForensics
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| RealForensicsTraining Dataset=FF++, Training Compression=c23, Weight Source=reproduced using officially provided weights, Model Type=Video-based2024.03 | 99.6 | 91.8 | |
| AltFreezeTraining Dataset=FF++, Training Compression=c23, Weight Source=reproduced using officially provided weights, Model Type=Video-based2024.03 | 99.2 | 95.2 | |
| QAD-RSetting=Quality-agnostic, Compression=Video (raw + c23 + c40)2023.09 | 99.12 | — | |
| QAD-ESetting=Quality-agnostic, Compression=Video (raw + c23 + c40)2023.09 | 99.12 | — | |
| OursTraining Dataset=FF++, Training Compression=c23, Model Type=Video-based2024.03 | 99 | 95.8 | |
| FTCNTraining Dataset=FF++, Training Compression=c23, Weight Source=reproduced using officially provided weights, Model Type=Video-based2024.03 | 98.8 | 94.7 | |
| Fang & LinSetting=Quality-agnostic, Compression=Video (raw + c23 + c40)2023.09 | 98.66 | — | |
| Rössler et al.Setting=Quality-agnostic, Compression=Video (raw + c23 + c40), Implementation source=published code2023.09 | 98.5 | — | |
| F3NetSetting=Quality-agnostic, Compression=Video (raw + c23 + c40), Implementation source=published code2023.09 | 97.45 | — | |
| MATSetting=Quality-agnostic, Compression=Video (raw + c23 + c40), Implementation source=published code2023.09 | 97.45 | — | |
| LipForensicsTraining Dataset=FF++, Training Compression=c23, Weight Source=obtained from AltFreeze [55], Model Type=Video-based2024.03 | 97.1 | — | |
| MesoNetSetting=Quality-agnostic, Compression=Video (raw + c23 + c40)2023.09 | 96 | — | |
| ADDSetting=Quality-agnostic, Compression=Video (raw + c23 + c40), Implementation source=pre-trained weights2023.09 | 95.94 | — | |
| Face X-rayTraining Dataset=FF++, Training Compression=c23, Weight Source=obtained from AltFreeze [55], Model Type=Image-based2024.03 | 92.8 | 88.6 | |
| StyleGRUTraining Dataset=FF++, Training Compression=c23, Stage=Stage 1 training, Model Type=Video-based2024.03 | 90.7 | 79.8 | |
| CNN-GRUTraining Dataset=FF++, Training Compression=c23, Weight Source=obtained from AltFreeze [55], Model Type=Video-based2024.03 | 80.8 | — | |
| SBIsSetting=Quality-agnostic, Compression=Video (raw + c23 + c40), Implementation source=pre-trained weights2023.09 | 80.48 | — | |
| SBI-R50Training Dataset=FF++, Training Compression=c23, Weight Source=obtained from AltFreeze [55], Backbone=ResNet-50, Model Type=Image-based2024.03 | 78.2 | 87.3 | |
| XceptionTraining Dataset=FF++, Training Compression=c23, Weight Source=obtained from AltFreeze [55], Model Type=Image-based2024.03 | 72 | — | |
| Multi-taskTraining Dataset=FF++, Training Compression=c23, Weight Source=obtained from AltFreeze [55], Model Type=Image-based2024.03 | 66 | — | |
| CNN-augTraining Dataset=FF++, Training Compression=c23, Weight Source=obtained from AltFreeze [55], Model Type=Image-based2024.03 | 65.7 | — | |
| FWATraining Dataset=FF++, Training Compression=c23, Weight Source=obtained from AltFreeze [55], Model Type=Image-based2024.03 | 65.5 | — | |
| PatchForensicsTraining Dataset=FF++, Training Compression=c23, Weight Source=obtained from AltFreeze [55], Model Type=Image-based2024.03 | 57.8 | — |