Deepfake Detection on DeepfakeDetection (DFD) (test)
97.56AUCSBIs
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SBIsTraining Dataset=FF++, Feature Extracting Backbone=EfficientNet-B42026.01 | 97.56 | — | |
| SBIsTraining Dataset=FF++, Feature Extracting Backbone=Swin Transformer2026.01 | 97.34 | — | |
| Ours†Training Dataset=FF++, Feature Extracting Backbone=Swin Transformer2026.01 | 97.01 | — | |
| Ours*Training Dataset=FF++, Feature Extracting Backbone=EfficientNet-B42026.01 | 96.23 | — | |
| OursTraining Dataset=FF++, Training Compression=c23, Model Type=Video-based2024.03 | 96.1 | 95.8 | |
| Face X-rayTraining Dataset=FF++, Training Compression=c23, Weight Source=obtained from AltFreeze [55], Model Type=Image-based2024.03 | 95.4 | 88.6 | |
| Face-X-RayTraining Dataset=FF++2026.01 | 95.4 | — | |
| CDFARetrained=false, TSRL Integration=false2026.03 | 95.4 | — | |
| FTCNTraining Dataset=FF++, Training Compression=c23, Weight Source=reproduced using officially provided weights, Model Type=Video-based2024.03 | 94.4 | 94.7 | |
| SBI-R50Training Dataset=FF++, Training Compression=c23, Weight Source=obtained from AltFreeze [55], Backbone=ResNet-50, Model Type=Image-based2024.03 | 94 | 87.3 | |
| AltFreezeTraining Dataset=FF++, Training Compression=c23, Weight Source=reproduced using officially provided weights, Model Type=Video-based2024.03 | 93.7 | 95.2 | |
| SRI-Net2026.02 | 93.1 | — | |
| DCLTraining Dataset=FF++, Training Compression=c23, Weight Source=obtained from AltFreeze [55], Model Type=Image-based2024.03 | 91.6 | — | |
| EffortRetrained=true, TSRL Integration=false2026.03 | 91 | — | |
| SLADDRetrained=false, TSRL Integration=false2026.03 | 90.4 | — | |
| Effort + TSRLRetrained=true, TSRL Integration=true2026.03 | 90.4 | — | |
| Xception2026.02 | 89.6 | — | |
| LRLTraining Dataset=FF++2026.01 | 89.24 | — | |
| RECCE2026.02 | 89.1 | — | |
| RECCERetrained=false, TSRL Integration=false2026.03 | 89.1 | — | |
| IID + TSRLRetrained=true, TSRL Integration=true2026.03 | 88.9 | — | |
| SRMRetrained=false, TSRL Integration=false2026.03 | 88.5 | — | |
| SBI2026.02 | 88.2 | — | |
| LSDARetrained=false, TSRL Integration=false2026.03 | 88.1 | — | |
| XceptionTraining Dataset=FF++2026.01 | 87.86 | — | |
| ProDetRetrained=true, TSRL Integration=false2026.03 | 87.8 | — | |
| IIDRetrained=true, TSRL Integration=false2026.03 | 87.6 | — | |
| SPSLRetrained=false, TSRL Integration=false2026.03 | 87.1 | — | |
| CORERetrained=true, TSRL Integration=false2026.03 | 86.8 | — | |
| UCF2026.02 | 86.7 | — | |
| CORE + TSRLRetrained=true, TSRL Integration=true2026.03 | 86.3 | — | |
| EffiNet-B42026.02 | 86.2 | — | |
| ProDet + TSRLRetrained=true, TSRL Integration=true2026.03 | 86.1 | — | |
| UCF + TSRLRetrained=true, TSRL Integration=true2026.03 | 85 | — | |
| UCFRetrained=true, TSRL Integration=false2026.03 | 84.5 | — | |
| F3-Net2026.02 | 84.4 | — | |
| F3NetRetrained=false, TSRL Integration=false2026.03 | 84.4 | — | |
| TALLRetrained=false, TSRL Integration=false2026.03 | 83.3 | — | |
| SBIRetrained=false, TSRL Integration=false2026.03 | 82.7 | — | |
| RealForensicsTraining Dataset=FF++, Training Compression=c23, Weight Source=reproduced using officially provided weights, Model Type=Video-based2024.03 | 82.2 | 91.8 | |
| CLIPRetrained=true, TSRL Integration=false2026.03 | 75.2 | — | |
| CLIP + TSRLRetrained=true, TSRL Integration=true2026.03 | 73.2 | — | |
| StyleGRUTraining Dataset=FF++, Training Compression=c23, Stage=Stage 1 training, Model Type=Video-based2024.03 | 72.2 | 79.8 |