Deepfake Detection on CDF v2
0.963AUCHAAD
Evaluation results
| Method | Links | |
|---|---|---|
| HAADTraining Dataset=FF++, Evaluation Protocol=Cross-dataset Evaluation2026.05 | 0.963 | |
| OursTraining Dataset=FaceForensics++ [48], Evaluation Protocol=cross-dataset2026.05 | 0.9575 | |
| OursDetector Type=Language detector2026.05 | 0.9575 | |
| EffortDetector Type=Binary detector2026.05 | 0.956 | |
| LAADetector Type=Binary detector2026.05 | 0.954 | |
| X^2-DFDTraining Dataset=FF++, Evaluation Protocol=Cross-dataset Evaluation, Re-implemented=true2026.05 | 0.951 | |
| M2F2Training Dataset=FaceForensics++ [48], Evaluation Protocol=cross-dataset2026.05 | 0.951 | |
| M2F2Detector Type=Language detector2026.05 | 0.951 | |
| UCLVLMTraining Dataset=FaceForensics++ [48], Evaluation Protocol=cross-dataset2026.05 | 0.9471 | |
| UCLVLMDetector Type=Language detector2026.05 | 0.9471 | |
| EffortTraining Dataset=FF++, Evaluation Protocol=Cross-dataset Evaluation, Re-implemented=true2026.05 | 0.945 | |
| BLIP-TI-SBITraining Dataset=FaceForensics++ [48], Evaluation Protocol=cross-dataset2026.05 | 0.9398 | |
| BLIP-TI-SBIDetector Type=Language detector2026.05 | 0.9398 | |
| VLFFDTraining Dataset=FF++, Evaluation Protocol=Cross-dataset Evaluation, Re-implemented=true2026.05 | 0.938 | |
| FIA-USATraining Dataset=FF++, Evaluation Protocol=Cross-dataset Evaluation, Re-implemented=true2026.05 | 0.935 | |
| SBIDetector Type=Binary detector2026.05 | 0.9318 | |
| FakeSTormerDetector Type=Binary detector2026.05 | 0.924 | |
| FakeRadarBackbone=Vision Transformer, Input Type=Video2025.12 | 0.917 | |
| PMMDetector Type=Binary detector2026.05 | 0.9153 | |
| DSMTraining Dataset=FF++ [7], Publication Year=20252026.03 | 0.9127 | |
| LSDABackbone=EfficientNet, Input Type=Frame2025.12 | 0.911 | |
| DAIDTraining Dataset=FF++, Evaluation Protocol=Cross-dataset Evaluation, Re-implemented=true2026.05 | 0.909 | |
| TALLBackbone=Swin Transformer, Input Type=Video2025.12 | 0.908 | |
| SpInShieldAttack Type=Spectral Tilt2026.05 | 0.9052 | |
| PCL+I2GBackbone=Resnet, Input Type=Frame2025.12 | 0.9 | |
| SpInShieldAttack Type=SNR-based Noise2026.05 | 0.8982 | |
| AltFreezingBackbone=3D ResNet, Input Type=Video2025.12 | 0.895 | |
| NACOBackbone=Vision Transformer, Input Type=Video2025.12 | 0.895 | |
| EFFTTraining Dataset=FF++ [7], Publication Year=20252026.03 | 0.8945 | |
| LTTDBackbone=Vision Transformer, Input Type=Video2025.12 | 0.893 | |
| StyleFlowBackbone=3D ResNet, Input Type=Video2025.12 | 0.89 | |
| SBITraining Dataset=FF++, Evaluation Protocol=Cross-dataset Evaluation2026.05 | 0.886 | |
| LSDATraining Dataset=FF++, Evaluation Protocol=Cross-dataset Evaluation2026.05 | 0.875 | |
| SeeABLEBackbone=EfficientNet, Input Type=Frame2025.12 | 0.873 | |
| FTCNBackbone=3D ResNet, Input Type=Video2025.12 | 0.869 | |
| RealForensicsBackbone=3D ResNet, Input Type=Video2025.12 | 0.869 | |
| HFRTraining Dataset=FF++ [7]2026.03 | 0.8534 | |
| Controllable GSDetector Type=Binary detector2026.05 | 0.8497 | |
| VL-FFDTraining Dataset=FaceForensics++ [48], Evaluation Protocol=cross-dataset2026.05 | 0.848 | |
| VL-FFDDetector Type=Language detector2026.05 | 0.848 | |
| SRMTraining Dataset=FF++, Evaluation Protocol=Cross-dataset Evaluation2026.05 | 0.84 | |
| IIDTraining Dataset=FF++, Evaluation Protocol=Cross-dataset Evaluation2026.05 | 0.838 | |
| UCFTraining Dataset=FF++, Evaluation Protocol=Cross-dataset Evaluation2026.05 | 0.837 | |
| GM-DFTraining Dataset=FF++ [7], Publication Year=20242026.03 | 0.8316 | |
| LSDATraining Dataset=FF++ [7], Publication Year=20242026.03 | 0.83 | |
| LSDADetector Type=Binary detector2026.05 | 0.83 | |
| UIA-ViTDetector Type=Binary detector2026.05 | 0.8241 | |
| UIA-ViTBackbone=Vision Transformer, Input Type=Frame2025.12 | 0.824 | |
| UCFBackbone=Xception, Input Type=Frame2025.12 | 0.824 | |
| RECCETraining Dataset=FF++, Evaluation Protocol=Cross-dataset Evaluation2026.05 | 0.823 | |
| CORETraining Dataset=FF++, Evaluation Protocol=Cross-dataset Evaluation2026.05 | 0.809 | |
| SPSLTraining Dataset=FF++, Evaluation Protocol=Cross-dataset Evaluation2026.05 | 0.799 | |
| SLADDBackbone=Xception, Input Type=Frame2025.12 | 0.797 | |
| SSLTraining Dataset=FF++ [7], Publication Year=20222026.03 | 0.797 | |
| Face X-RayBackbone=HRNet, Input Type=Frame2025.12 | 0.795 | |
| FaceXRay+BIDetector Type=Binary detector2026.05 | 0.795 | |
| F3NetTraining Dataset=FF++, Evaluation Protocol=Cross-dataset Evaluation2026.05 | 0.789 | |
| SeeABLEDetector Type=Binary detector2026.05 | 0.783 | |
| SPSLDetector Type=Binary detector2026.05 | 0.7688 | |
| SFGDTraining Dataset=FF++ [7], Publication Year=20232026.03 | 0.7583 | |
| UCFTraining Dataset=FF++ [7], Publication Year=20232026.03 | 0.7527 | |
| UCFDetector Type=Binary detector2026.05 | 0.7527 | |
| EfficientB4Training Dataset=FF++ [7], Publication Year=20232026.03 | 0.7487 | |
| EfficientNet-B4Detector Type=Binary detector2026.05 | 0.7487 | |
| D3Training Dataset=FF++ [7], Publication Year=20252026.03 | 0.7455 | |
| PFGTraining Dataset=FF++ [7], Publication Year=20242026.03 | 0.7442 | |
| CORETraining Dataset=FF++ [7], Publication Year=20222026.03 | 0.7428 | |
| XceptionBackbone=Xception, Input Type=Frame2025.12 | 0.737 | |
| XceptionTraining Dataset=FF++ [7], Publication Year=20232026.03 | 0.7365 | |
| F3NetTraining Dataset=FF++ [7], Publication Year=20232026.03 | 0.7352 | |
| RecceTraining Dataset=FF++ [7], Publication Year=20222026.03 | 0.7319 | |
| BLIP-TI-HiFiDetector Type=Language detector2026.05 | 0.71 | |
| BLIP-TI-RECCEDetector Type=Language detector2026.05 | 0.7021 | |
| RECCEDetector Type=Binary detector2026.05 | 0.6871 | |
| DFD-FCGAttack Type=Spectral Tilt2026.05 | 0.6852 | |
| FakeSTAttack Type=Spectral Tilt2026.05 | 0.6524 | |
| DFD-FCGAttack Type=SNR-based Noise2026.05 | 0.6475 | |
| SLFAttack Type=Spectral Tilt2026.05 | 0.6455 | |
| NACOAttack Type=Spectral Tilt2026.05 | 0.642 | |
| PTFDAttack Type=Spectral Tilt2026.05 | 0.6385 | |
| PBDTraining Dataset=FF++ [7], Publication Year=20212026.03 | 0.633 | |
| FTCNAttack Type=Spectral Tilt2026.05 | 0.6252 | |
| AltFreAttack Type=Spectral Tilt2026.05 | 0.6244 | |
| BLIP-TI-Xcep.Detector Type=Language detector2026.05 | 0.6241 | |
| FakeSTAttack Type=SNR-based Noise2026.05 | 0.6238 | |
| NACOAttack Type=SNR-based Noise2026.05 | 0.6215 | |
| SLFAttack Type=SNR-based Noise2026.05 | 0.6188 | |
| TALLAttack Type=Spectral Tilt2026.05 | 0.6155 | |
| PTFDAttack Type=SNR-based Noise2026.05 | 0.6122 | |
| XceptionDetector Type=Binary detector2026.05 | 0.6118 | |
| FTCNAttack Type=SNR-based Noise2026.05 | 0.604 | |
| AltFreAttack Type=SNR-based Noise2026.05 | 0.6015 | |
| TALLAttack Type=SNR-based Noise2026.05 | 0.5855 | |
| AFWAttack Type=Spectral Tilt2026.05 | 0.5852 | |
| VideoMAE(B)Attack Type=Spectral Tilt2026.05 | 0.5743 | |
| AFWAttack Type=SNR-based Noise2026.05 | 0.5645 | |
| VideoMAE(B)Attack Type=SNR-based Noise2026.05 | 0.552 |