Fake Face Detection on Celeb-DF v2 (test)
99.97AUCPatch + RFM
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Patch + RFMBackbone=Patch2021.04 | 99.97 | 93.44 | 89.58 | — | |
| PatchBackbone=Patch2021.04 | 99.96 | 91.83 | 86.16 | — | |
| HuForDet2026.01 | 99.96 | — | — | 99.01 | |
| Xception + RFMBackbone=Xception2021.04 | 99.94 | 93.88 | 87.08 | — | |
| RECCE2026.01 | 99.94 | — | — | 98.59 | |
| Xception + REBackbone=Xception2021.04 | 99.89 | 88.11 | 85.2 | — | |
| XceptionBackbone=Xception2021.04 | 99.85 | 89.11 | 84.22 | — | |
| Xception + AEBackbone=Xception2021.04 | 99.84 | 84.05 | 76.63 | — | |
| Add-Net2026.01 | 99.55 | — | — | 96.93 | |
| M2F22026.01 | 99.02 | — | — | 98.98 | |
| F3-Net2026.01 | 98.93 | — | — | 95.95 | |
| TALL2026.01 | 98.55 | — | — | 97.57 | |
| QAD-ESetting=Quality-agnostic, Compression=Video (raw + c23 + c40)2023.09 | 98.38 | — | — | — | |
| QAD-RSetting=Quality-agnostic, Compression=Video (raw + c23 + c40)2023.09 | 98.36 | — | — | — | |
| Rössler et al.Setting=Quality-agnostic, Compression=Video (raw + c23 + c40), Implementation source=published code2023.09 | 97.49 | — | — | — | |
| Fang & LinSetting=Quality-agnostic, Compression=Video (raw + c23 + c40)2023.09 | 96.58 | — | — | — | |
| F3NetSetting=Quality-agnostic, Compression=Video (raw + c23 + c40), Implementation source=published code2023.09 | 95.06 | — | — | — | |
| MATSetting=Quality-agnostic, Compression=Video (raw + c23 + c40), Implementation source=published code2023.09 | 95.06 | — | — | — | |
| EFNB4 + SBIsInput Type=Frame, Training Set (Real)=true, Training Set (Fake)=false2022.04 | 93.18 | — | — | — | |
| PCL + I2GInput Type=Frame, Training Set (Real)=true, Training Set (Fake)=true2022.04 | 90.03 | — | — | — | |
| PCL+I2Gtrained_on=FF++, evaluation_level=video-level2022.07 | 90 | — | — | — | |
| SBI+EB4trained_on=FF++, evaluation_level=video-level2022.07 | 89.9 | — | — | — | |
| LTTDtrained_on=FF++, evaluation_level=video-level2022.07 | 89.3 | — | — | — | |
| FTCNInput Type=Video, Training Set (Real)=true, Training Set (Fake)=true2022.04 | 86.9 | — | — | — | |
| FTCN-TTtrained_on=FF++, evaluation_level=video-level2022.07 | 86.9 | — | — | — | |
| LipForensicsInput Type=Video, Training Set (Real)=true, Training Set (Fake)=true2022.04 | 82.4 | — | — | — | |
| LipForTrain Set=FF++(c23)2022.09 | 82.4 | — | — | — | |
| LipForensicstrained_on=FF++, evaluation_level=video-level2022.07 | 82.4 | — | — | — | |
| MesoNetSetting=Quality-agnostic, Compression=Video (raw + c23 + c40)2023.09 | 80.52 | — | — | — | |
| Face X-raytrained_on=FF++, evaluation_level=video-level2022.07 | 79.5 | — | — | — | |
| FRDMInput Type=Frame, Training Set (Real)=true, Training Set (Fake)=true2022.04 | 79.4 | — | — | — | |
| LRLInput Type=Frame, Training Set (Real)=true, Training Set (Fake)=true2022.04 | 78.26 | — | — | — | |
| LRLTrain Set=FF++(c23)2022.09 | 78.26 | — | — | — | |
| SPSLTrain Set=FF++(c23)2022.09 | 76.88 | — | — | — | |
| Two-branchInput Type=Video, Training Set (Real)=true, Training Set (Fake)=true2022.04 | 76.65 | — | — | — | |
| SOLA -supTrain Set=DF(c23), supervision=supervised2022.09 | 76.02 | — | — | — | |
| Multi-tasktrained_on=FF++, evaluation_level=video-level2022.07 | 75.5 | — | — | — | |
| DAMInput Type=Video, Training Set (Real)=true, Training Set (Fake)=true2022.04 | 75.3 | — | — | — | |
| Two BranchTrain Set=FF++(c40)2022.09 | 73.41 | — | — | — | |
| SOLA -weakly supTrain Set=DF(c23), supervision=weakly supervised2022.09 | 72.47 | — | — | — | |
| MTD-NetTrain Set=FF++(c23)2022.09 | 70.12 | — | — | — | |
| CNN-GRUtrained_on=FF++, evaluation_level=video-level2022.07 | 69.8 | — | — | — | |
| PatchForensicstrained_on=FF++, evaluation_level=video-level2022.07 | 69.6 | — | — | — | |
| FWAtrained_on=FF++, evaluation_level=video-level2022.07 | 69.5 | — | — | — | |
| DSP-FWAInput Type=Frame, Training Set (Real)=true, Training Set (Fake)=false2022.04 | 69.3 | — | — | — | |
| MADDTrain Set=FF++(c23)2022.09 | 67.44 | — | — | — | |
| F3NetTrain Set=FF++(c23)2022.09 | 65.2 | — | — | — | |
| GFFTrain Set=FF++(c23)2022.09 | 65.2 | — | — | — | |
| F³ NetTrain Set=FF++(c23)2022.09 | 65.17 | — | — | — | |
| FWATrain Set=Self-made2022.09 | 57.32 | — | — | — |