Deepfake Detection on WildDeepfake
0.889AUCAletheia
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Aletheia2026.04 | 0.889 | — | |
| DF40 baseline2026.04 | 0.853 | — | |
| EffortTraining Dataset=FaceForensics++ (c23), Evaluation Protocol=Cross-dataset2026.04 | 0.848 | — | |
| LAA-Net2026.04 | 0.841 | — | |
| SePLTraining Dataset=FaceForensics++ (c23), Evaluation Protocol=Cross-dataset2026.04 | 0.838 | — | |
| Face-D2CLProtocol=Protocol 12026.04 | 0.812 | 0.876 | |
| Xception baseline2026.04 | 0.812 | — | |
| ForAdaTraining Dataset=FaceForensics++ (c23), Evaluation Protocol=Cross-dataset2026.04 | 0.805 | — | |
| LSDATraining Dataset=FaceForensics++ (c23), Evaluation Protocol=Cross-dataset2026.04 | 0.797 | — | |
| CDFATraining Dataset=FaceForensics++ (c23), Evaluation Protocol=Cross-dataset2026.04 | 0.796 | — | |
| ProDetTraining Dataset=FaceForensics++ (c23), Evaluation Protocol=Cross-dataset2026.04 | 0.781 | — | |
| UCFTraining Dataset=FaceForensics++ (c23), Evaluation Protocol=Cross-dataset2026.04 | 0.774 | — | |
| RECCETraining Dataset=FaceForensics++ (c23), Evaluation Protocol=Cross-dataset2026.04 | 0.756 | — | |
| F3NetTraining Dataset=FaceForensics++ (c23), Evaluation Protocol=Cross-dataset2026.04 | 0.728 | — | |
| CORETraining Dataset=FaceForensics++ (c23), Evaluation Protocol=Cross-dataset2026.04 | 0.724 | — | |
| DFILProtocol=Protocol 12026.04 | 0.708 | 0.783 | |
| SAIDOProtocol=Protocol 12026.04 | 0.705 | 0.797 | |
| SBITraining Dataset=FaceForensics++ (c23), Evaluation Protocol=Cross-dataset2026.04 | 0.703 | — | |
| SPSLTraining Dataset=FaceForensics++ (c23), Evaluation Protocol=Cross-dataset2026.04 | 0.702 | — | |
| SRMTraining Dataset=FaceForensics++ (c23), Evaluation Protocol=Cross-dataset2026.04 | 0.702 | — | |
| SUR-LIDProtocol=Protocol 12026.04 | 0.692 | 0.779 | |
| SLADDTraining Dataset=FaceForensics++ (c23), Evaluation Protocol=Cross-dataset2026.04 | 0.69 | — | |
| DFFreqProtocol=Protocol 12026.04 | 0.681 | 0.665 | |
| DFD-FCGProtocol=Protocol 12026.04 | 0.673 | 0.724 | |
| IIDTraining Dataset=FaceForensics++ (c23), Evaluation Protocol=Cross-dataset2026.04 | 0.666 | — |