Deepfake Detection on DFD cross-domain
0.36Gender FFPRDAG-FDD
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| DAG-FDDBackbone=Xception2025.11 | 0.36 | 11.35 | 65.12 | 13.02 | 10.39 | 63.57 | 27.3 | 24.24 | 57.27 | 68.2 | |
| OursBackbone=ResNet-502025.11 | 0.61 | 0.88 | 63.24 | 1.08 | 13.29 | 54.91 | 12.2 | 15.93 | 45.08 | 64.35 | |
| DAG-FDDBackbone=ResNet-502025.11 | 5.22 | 6.85 | 57.48 | 6.69 | 17.95 | 50.66 | 34.71 | 26.31 | 46.82 | 58.75 | |
| Synergistic Fairness Optimization frameworkBackbone=Xception2025.11 | 6.25 | 10.94 | 68.61 | 12.33 | 6.09 | 66.87 | 19.26 | 12 | 60.95 | 70.83 | |
| OriBackbone=Xception2025.11 | 7.81 | 11.44 | 61.91 | 13.28 | 6.19 | 62.11 | 48.35 | 17.02 | 45.47 | 67.51 | |
| OriBackbone=ResNet-502025.11 | 11.89 | 7.88 | 66.2 | 11.07 | 24.81 | 53.77 | 73.7 | 43.89 | 46.76 | 63.76 | |
| DAW-FDDBackbone=Xception2025.11 | 14.89 | 13.39 | 50.1 | 14.67 | 15.89 | 47.04 | 45.36 | 37.8 | 44.58 | 53.71 | |
| FairadapterBackbone=ViT-L/142025.11 | 15.12 | 12.4 | 68.17 | 28.18 | 19.26 | 54.72 | 48.29 | 38 | 57.65 | 60.78 | |
| FairadapterBackbone=ViT-L/142025.11 | 15.12 | 12.4 | 68.17 | 28.18 | 9.26 | 65.72 | 48.29 | 38 | 57.65 | 70.78 | |
| DAW-FDDBackbone=ResNet-502025.11 | 18.99 | 18.81 | 60.99 | 4.02 | 22.74 | 52.84 | 50.34 | 45.11 | 49.43 | 64.03 |