Deepfake Detection on DeeperForensics (DFo)
99.6Video-level AUCFCG
Evaluation results
| Method | Links | ||
|---|---|---|---|
| FCGPristine only=false2025.12 | 99.6 | — | |
| PTFPristine only=false2025.12 | 99.4 | — | |
| AltfreezingPristine only=false2025.12 | 99.3 | — | |
| SFAPristine only=false2025.12 | 99.24 | — | |
| VBPristine only=true, Model selection protocol (test-time data)=true2025.12 | 99.1 | — | |
| KiMoIPristine only=true2025.12 | 98.47 | — | |
| EffortPristine only=false, Model selection protocol (test-time data)=true2025.12 | 97.7 | — | |
| CDFAPristine only=false2025.12 | 97.3 | — | |
| LSDAPristine only=false, Model selection protocol (test-time data)=true2025.12 | 89.2 | — | |
| SBI (c23)Pristine only=true2025.12 | 84.36 | — | |
| UCFPristine only=false2025.12 | 80.8 | — | |
| SLADDPristine only=false2025.12 | 80 | — | |
| RECCEPristine only=false2025.12 | 78.4 | — | |
| F3NetPristine only=false2025.12 | 73 | — | |
| SPSLPristine only=false2025.12 | 72.3 | — | |
| SRMPristine only=false2025.12 | 72.2 | — | |
| CNN-augTraining Dataset=FaceForensics++, Evaluation Protocol=Cross-dataset generalization2022.01 | — | 74.4 | |
| CNN-GRUTraining Dataset=FaceForensics++, Evaluation Protocol=Cross-dataset generalization2022.01 | — | 74.1 | |
| CSNTraining Dataset=FaceForensics++, Evaluation Protocol=Cross-dataset generalization2022.01 | — | 89.3 | |
| DSP-FWATraining Dataset=FaceForensics++, Evaluation Protocol=Cross-dataset generalization2022.01 | — | 50.2 | |
| Face X-rayTraining Dataset=FaceForensics++, Evaluation Protocol=Cross-dataset generalization2022.01 | — | 86.8 | |
| FTCNTraining Dataset=FaceForensics++, Evaluation Protocol=Cross-dataset generalization2022.01 | — | 98.8 | |
| LipForensicsTraining Dataset=FaceForensics++, Evaluation Protocol=Cross-dataset generalization2022.01 | — | 97.6 | |
| Multi-taskTraining Dataset=FaceForensics++, Evaluation Protocol=Cross-dataset generalization2022.01 | — | 77.7 | |
| Patch-basedTraining Dataset=FaceForensics++, Evaluation Protocol=Cross-dataset generalization2022.01 | — | 81.8 | |
| RealForensicsTraining Dataset=FaceForensics++, Evaluation Protocol=Cross-dataset generalization2022.01 | — | 99.3 | |
| XceptionTraining Dataset=FaceForensics++, Evaluation Protocol=Cross-dataset generalization2022.01 | — | 84.5 |