Deepfake Detection on DFDCP (test)
96.6AUCSFAM
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SFAMTraining Dataset=UADFV2026.04 | 96.6 | — | |
| XceptionTraining Dataset=UADFV2026.04 | 94.4 | — | |
| ForensicsAdapterLearning Type=Pseudo-Supervised, Modality=Frame, Usage of Reference=false2026.01 | 93.98 | — | |
| ExposeAnyoneLearning Type=Self-Supervised, Modality=Audio & Video, Usage of Reference=true2026.01 | 93.45 | — | |
| SFAMTraining Dataset=CDv12026.04 | 93.2 | — | |
| ExposeAnyone w/ VoxCeleb2Learning Type=Self-Supervised, Modality=Audio & Video, Usage of Reference=true2026.01 | 93.06 | — | |
| EFFORTLearning Type=Supervised, Modality=Frame, Usage of Reference=false2026.01 | 92.89 | — | |
| Effort + TSRLRetrained=true, TSRL Integration=true2026.03 | 92.4 | — | |
| DFD-FCGLearning Type=Supervised, Modality=Video, Usage of Reference=false2026.01 | 90.04 | — | |
| EffortRetrained=true, TSRL Integration=false2026.03 | 89.9 | — | |
| SFAMTraining Dataset=CDv22026.04 | 89.2 | — | |
| LAA-Net w/ SBILearning Type=Pseudo-Supervised, Modality=Frame, Usage of Reference=false2026.01 | 88.96 | — | |
| SBILearning Type=Pseudo-Supervised, Modality=Frame, Usage of Reference=false2026.01 | 88.51 | — | |
| CDFARetrained=false, TSRL Integration=false2026.03 | 88.1 | — | |
| FSFMLearning Type=Supervised, Modality=Frame, Usage of Reference=false2026.01 | 86.94 | — | |
| SFAMTraining Dataset=DFDC2026.04 | 85.4 | — | |
| POI-ForensicsLearning Type=Self-Supervised, Modality=Audio & Video, Usage of Reference=true2026.01 | 85.34 | — | |
| SBIRetrained=false, TSRL Integration=false2026.03 | 84.8 | — | |
| ProDet + TSRLRetrained=true, TSRL Integration=true2026.03 | 84.5 | — | |
| ID-RevealLearning Type=Self-Supervised, Modality=Video, Usage of Reference=true2026.01 | 83.84 | — | |
| XceptionTraining Dataset=CDv12026.04 | 81.8 | — | |
| IIDTraining dataset=FF++ (C23)2026.02 | 81.23 | — | |
| LSDARetrained=false, TSRL Integration=false2026.03 | 81.2 | — | |
| Face X-rayTraining dataset=FF++ (C23)2026.02 | 80.92 | — | |
| UCFTraining dataset=FF++ (C23)2026.02 | 80.51 | — | |
| MSBA-CLIPTraining dataset=FF++ (C23)2026.02 | 80.38 | — | |
| ProDetRetrained=true, TSRL Integration=false2026.03 | 80.1 | — | |
| XceptionTraining Dataset=CDv22026.04 | 77.9 | — | |
| UCFLearning Type=Supervised, Modality=Frame, Usage of Reference=false2026.01 | 77.1 | — | |
| SPSLRetrained=false, TSRL Integration=false2026.03 | 77 | — | |
| CORERetrained=true, TSRL Integration=false2026.03 | 75.9 | — | |
| SLADDRetrained=false, TSRL Integration=false2026.03 | 75.6 | — | |
| IID + TSRLRetrained=true, TSRL Integration=true2026.03 | 75.2 | — | |
| F3NetRetrained=false, TSRL Integration=false2026.03 | 74.9 | — | |
| XceptionTraining Dataset=DFDC2026.04 | 74.6 | — | |
| ICTLearning Type=Pseudo-Supervised, Modality=Frame, Usage of Reference=false2026.01 | 74.59 | — | |
| TALLRetrained=false, TSRL Integration=false2026.03 | 73.9 | — | |
| RECCERetrained=false, TSRL Integration=false2026.03 | 73.4 | — | |
| SRMRetrained=false, TSRL Integration=false2026.03 | 72.8 | — | |
| CLIP + TSRLRetrained=true, TSRL Integration=true2026.03 | 72.4 | — | |
| CORE + TSRLRetrained=true, TSRL Integration=true2026.03 | 72.4 | — | |
| RealForensicsLearning Type=Supervised, Modality=Video, Usage of Reference=false2026.01 | 72.23 | — | |
| CORETraining dataset=FF++ (C23)2026.02 | 71.41 | — | |
| IIDRetrained=true, TSRL Integration=false2026.03 | 70.6 | — | |
| LipForensicsLearning Type=Supervised, Modality=Video, Usage of Reference=false2026.01 | 69.89 | — | |
| UCF + TSRLRetrained=true, TSRL Integration=true2026.03 | 69.8 | — | |
| AltFreezingLearning Type=Supervised, Modality=Video, Usage of Reference=false2026.01 | 69.33 | — | |
| UCFRetrained=true, TSRL Integration=false2026.03 | 69 | — | |
| Face X-rayLearning Type=Pseudo-Supervised, Modality=Frame, Usage of Reference=false2026.01 | 67.77 | — | |
| CLIPRetrained=true, TSRL Integration=false2026.03 | 66.7 | — | |
| SpeechForensicsLearning Type=Self-Supervised, Modality=Audio & Video, Usage of Reference=false2026.01 | 63.61 | — | |
| FTCNLearning Type=Supervised, Modality=Video, Usage of Reference=false2026.01 | 61.19 | — | |
| EfficientNet-b4Learning Type=Supervised, Modality=Frame, Usage of Reference=false2026.01 | 60.15 | — | |
| LipFDLearning Type=Supervised, Modality=Audio & Video, Usage of Reference=false2026.01 | 56.13 | — | |
| RECCELearning Type=Supervised, Modality=Frame, Usage of Reference=false2026.01 | 49.76 | — | |
| AVADLearning Type=Self-Supervised, Modality=Audio & Video, Usage of Reference=false2026.01 | 45.58 | — | |
| AdalogDetector=DeiT-Tiny, Type=Backbones2026.04 | — | 44.7 | |
| AdalogDetector=DeiT-Small, Type=Backbones2026.04 | — | 52.9 | |
| AdalogDetector=DeiT-Base, Type=Backbones2026.04 | — | 55.3 | |
| AdalogDetector=ViT-Small, Type=Backbones2026.04 | — | 55.4 | |
| AdalogDetector=ViT-Base, Type=Backbones2026.04 | — | 57.7 | |
| AdalogDetector=Swin-Small, Type=Backbones2026.04 | — | 56.7 | |
| AdalogDetector=Swin-Base, Type=Backbones2026.04 | — | 58.2 | |
| AdalogDetector=LSDA, Type=CVPR'242026.04 | — | 58.5 | |
| AdalogDetector=NPR, Type=CVPR'242026.04 | — | 59 | |
| AdalogDetector=UCF, Type=ICCV'232026.04 | — | 56.5 | |
| AdalogDetector=Wavelet, Type=WACV'252026.04 | — | 50.5 | |
| BRECQDetector=DeiT-Tiny, Type=Backbones2026.04 | — | 43.3 | |
| BRECQDetector=DeiT-Small, Type=Backbones2026.04 | — | 44.1 | |
| BRECQDetector=DeiT-Base, Type=Backbones2026.04 | — | 46.1 | |
| BRECQDetector=ViT-Small, Type=Backbones2026.04 | — | 46.2 | |
| BRECQDetector=ViT-Base, Type=Backbones2026.04 | — | 48.1 | |
| BRECQDetector=Swin-Small, Type=Backbones2026.04 | — | 47.2 | |
| BRECQDetector=Swin-Base, Type=Backbones2026.04 | — | 48.5 | |
| BRECQDetector=LSDA, Type=CVPR'242026.04 | — | 48.8 | |
| BRECQDetector=NPR, Type=CVPR'242026.04 | — | 49.2 | |
| BRECQDetector=UCF, Type=ICCV'232026.04 | — | 47.1 | |
| BRECQDetector=Wavelet, Type=WACV'252026.04 | — | 42.1 | |
| DeFakeQDetector=DeiT-Tiny, Type=Backbones2026.04 | — | 63.7 | |
| DeFakeQDetector=DeiT-Small, Type=Backbones2026.04 | — | 66.4 | |
| DeFakeQDetector=DeiT-Base, Type=Backbones2026.04 | — | 69.2 | |
| DeFakeQDetector=ViT-Small, Type=Backbones2026.04 | — | 71.6 | |
| DeFakeQDetector=ViT-Base, Type=Backbones2026.04 | — | 72.3 | |
| DeFakeQDetector=Swin-Small, Type=Backbones2026.04 | — | 70.5 | |
| DeFakeQDetector=Swin-Base, Type=Backbones2026.04 | — | 71.9 | |
| DeFakeQDetector=LSDA, Type=CVPR'242026.04 | — | 71.5 | |
| DeFakeQDetector=NPR, Type=CVPR'242026.04 | — | 72.1 | |
| DeFakeQDetector=UCF, Type=ICCV'232026.04 | — | 70.2 | |
| DeFakeQDetector=Wavelet, Type=WACV'252026.04 | — | 72.1 | |
| FIMA-QDetector=DeiT-Tiny, Type=Backbones2026.04 | — | 54.1 | |
| FIMA-QDetector=DeiT-Small, Type=Backbones2026.04 | — | 55.1 | |
| FIMA-QDetector=DeiT-Base, Type=Backbones2026.04 | — | 57.6 | |
| FIMA-QDetector=ViT-Small, Type=Backbones2026.04 | — | 57.8 | |
| FIMA-QDetector=ViT-Base, Type=Backbones2026.04 | — | 60.2 | |
| FIMA-QDetector=Swin-Small, Type=Backbones2026.04 | — | 59 | |
| FIMA-QDetector=Swin-Base, Type=Backbones2026.04 | — | 60.7 | |
| FIMA-QDetector=LSDA, Type=CVPR'242026.04 | — | 61 | |
| FIMA-QDetector=NPR, Type=CVPR'242026.04 | — | 61.5 | |
| FIMA-QDetector=UCF, Type=ICCV'232026.04 | — | 58.9 | |
| FIMA-QDetector=Wavelet, Type=WACV'252026.04 | — | 52.6 |