Deepfake Detection on DFD (Demographic Subgroup Metrics)
4.72Gender FPR (F)Ours
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| OursBackbone=Xception2025.11 | 4.72 | 5.49 | 78.98 | 2.09 | 18.66 | 77.44 | 27.32 | 22.81 | 69.73 | 81.46 | |
| PG-FDDBackbone=Xception2025.11 | 5.05 | 1.86 | 76.37 | 5.79 | 19.31 | 74.81 | 28 | 23.93 | 66.32 | 80.7 | |
| DAW-FDDBackbone=Xception2025.11 | 5.3 | 10.63 | 71.68 | 4.32 | 20.16 | 63.24 | 33.81 | 24.72 | 56.01 | 73.71 | |
| DAG-FDDBackbone=Xception2025.11 | 5.41 | 5.6 | 75.87 | 2.56 | 21.34 | 70.14 | 33 | 23.02 | 61.09 | 76.29 | |
| FairadapterBackbone=ViT-L/142025.11 | 6.32 | 7.55 | 63.28 | 11.29 | 12.28 | 56.28 | 29.56 | 33.11 | 48.63 | 68.12 | |
| OriBackbone=Xception2025.11 | 9.45 | 6.63 | 72.68 | 7.75 | 22.31 | 69.07 | 35.06 | 25.53 | 59.2 | 74.32 | |
| RSEF-FDDBackbone=Xception2025.11 | 18.5 | 6.11 | 75.97 | 19.45 | 19.4 | 72.65 | 35.34 | 10.97 | 67.24 | 80.54 |