Face Verification on RFW (Calibration and Group Fairness)
99.51Min-Group AUROCAC-Ridge
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| AC-RidgeBackbone=ArcFace R50, Dem-aware (require group labels at inference)=false2026.06 | 99.51 | 99.71 | 94.11 | 97.19 | 2.63 | 1.65 | 4 | — | — | — | |
| AC-MLPBackbone=ArcFace R50, Dem-aware (require group labels at inference)=false2026.06 | 98.93 | 99.33 | 91.58 | 96.54 | 2.14 | 1.88 | 0 | — | — | — | |
| AC-LinearBackbone=ArcFace R50, Dem-aware (require group labels at inference)=false2026.06 | 98.89 | 99.31 | 93.49 | 96.82 | 2.05 | 1.74 | 2 | — | — | — | |
| BaselineBackbone=ArcFace R50, Dem-aware (require group labels at inference)=false2026.06 | 98.83 | 99.25 | 93.14 | 96.58 | 27.97 | 12.86 | — | — | — | — | |
| FALCONBackbone=ArcFace R50, Dem-aware (require group labels at inference)=false2026.06 | 98.83 | 99.25 | 93.14 | 96.64 | 32.36 | 14.06 | 2 | — | — | — | |
| OracleBackbone=ArcFace R50, Dem-aware (require group labels at inference)=true2026.06 | 98.83 | 99.26 | 93.94 | 96.8 | 1.07 | 16.7 | — | — | — | — | |
| GSTBackbone=ArcFace R50, Dem-aware (require group labels at inference)=true2026.06 | 98.83 | 99.24 | 94.04 | 96.77 | 1.11 | 1.7 | — | — | — | — | |
| DemoNormBackbone=ArcFace R50, Dem-aware (require group labels at inference)=true2026.06 | 98.83 | 99.24 | 94.02 | 96.72 | 15.64 | 12.98 | — | — | — | — | |
| FairCalBackbone=ArcFace R50, Dem-aware (require group labels at inference)=false2026.06 | 98.81 | 99.26 | 93.49 | 96.77 | 1.02 | 1.7 | 3 | — | — | — | |
| FRAPPEBackbone=ArcFace R50, Dem-aware (require group labels at inference)=false2026.06 | 98.78 | 99.2 | 91.8 | 96.01 | 1.08 | 1.96 | — | — | — | — | |
| FSNBackbone=ArcFace R50, Dem-aware (require group labels at inference)=false2026.06 | 98.67 | 99.13 | 85.36 | 94.57 | 27.42 | 12.02 | 0 | — | — | — | |
| AC-DensityBackbone=AdaFace, Demographic-aware=false2026.06 | 98.2 | — | — | — | — | — | — | — | — | — | |
| AC-LinearBackbone=AdaFace, Demographic-aware=false2026.06 | 98 | — | — | — | — | — | — | — | — | — | |
| AC-MLPBackbone=AdaFace, Demographic-aware=false2026.06 | 97.9 | — | — | — | — | — | — | — | — | — | |
| FairCalBackbone=AdaFace, Demographic-aware=false2026.06 | 97.7 | — | — | — | — | — | — | — | — | — | |
| BaselineBackbone=AdaFace, Demographic-aware=false2026.06 | 97.6 | — | — | — | — | — | — | — | — | — | |
| FALCONBackbone=AdaFace, Demographic-aware=false2026.06 | 97.6 | — | — | — | — | — | — | — | — | — | |
| GSTBackbone=AdaFace, Demographic-aware=true2026.06 | 97.6 | — | — | — | — | — | — | — | — | — | |
| DemoNormBackbone=AdaFace, Demographic-aware=true2026.06 | 97.6 | — | — | — | — | — | — | — | — | — | |
| OracleBackbone=AdaFace, Demographic-aware=true2026.06 | 97.6 | — | — | — | — | — | — | — | — | — | |
| FRAPPÉBackbone=AdaFace, Demographic-aware=false2026.06 | 97.5 | — | — | — | — | — | — | — | — | — | |
| AC-DensityBackbone=R100, Demographic-aware=false2026.06 | 97.4 | — | — | — | — | — | — | — | — | — | |
| FSNBackbone=AdaFace, Demographic-aware=false2026.06 | 97.1 | — | — | — | — | — | — | — | — | — | |
| AC-LinearBackbone=R100, Demographic-aware=false2026.06 | 97 | — | — | — | — | — | — | — | — | — | |
| AC-MLPBackbone=R100, Demographic-aware=false2026.06 | 96.9 | — | — | — | — | — | — | — | — | — | |
| FairCalBackbone=R100, Demographic-aware=false2026.06 | 96.7 | — | — | — | — | — | — | — | — | — | |
| BaselineBackbone=R100, Demographic-aware=false2026.06 | 96.3 | — | — | — | — | — | — | — | — | — | |
| FALCONBackbone=R100, Demographic-aware=false2026.06 | 96.3 | — | — | — | — | — | — | — | — | — | |
| GSTBackbone=R100, Demographic-aware=true2026.06 | 96.3 | — | — | — | — | — | — | — | — | — | |
| DemoNormBackbone=R100, Demographic-aware=true2026.06 | 96.3 | — | — | — | — | — | — | — | — | — | |
| OracleBackbone=R100, Demographic-aware=true2026.06 | 96.3 | — | — | — | — | — | — | — | — | — | |
| FRAPPÉBackbone=R100, Demographic-aware=false2026.06 | 95.9 | — | — | — | — | — | — | — | — | — | |
| AC-DensityBackbone=MagFace, Demographic-aware=false2026.06 | 95.9 | — | — | — | — | — | — | — | — | — | |
| AC-RidgeBackbone=ArcFace R50, demographic_labels_at_inference=false2026.06 | 95.87 | 97.35 | 44.57 | 75.84 | 4.23 | 6.38 | 1 | — | 43 | 1.57 | |
| FSNBackbone=R100, Demographic-aware=false2026.06 | 95.8 | — | — | — | — | — | — | — | — | — | |
| AC-LinearBackbone=MagFace, Demographic-aware=false2026.06 | 94.8 | — | — | — | — | — | — | — | — | — | |
| AC-LinearBackbone=ArcFace R50, demographic_labels_at_inference=false2026.06 | 94.78 | 96.43 | 49.6 | 69.65 | 2.04 | 7.07 | 2 | — | 43 | 1.53 | |
| AC-MLPBackbone=ArcFace R50, demographic_labels_at_inference=false2026.06 | 94.65 | 96.3 | 44.5 | 68.24 | 2.77 | 7.29 | 1 | — | 43 | 1.5 | |
| AC-MLPBackbone=MagFace, Demographic-aware=false2026.06 | 94.6 | — | — | — | — | — | — | — | — | — | |
| FairCalBackbone=MagFace, Demographic-aware=false2026.06 | 94.5 | — | — | — | — | — | — | — | — | — | |
| FairCalBackbone=ArcFace R50, demographic_labels_at_inference=false2026.06 | 94.49 | 96.13 | 51.24 | 68.92 | 1.94 | 7.37 | 1 | — | 43 | 1.3 | |
| FALCONBackbone=ArcFace R50, demographic_labels_at_inference=false2026.06 | 94.13 | 95.69 | 50.02 | 66.68 | 27.85 | 18.82 | 4 | — | 53 | 1.57 | |
| BaselineBackbone=MagFace, Demographic-aware=false2026.06 | 94.1 | — | — | — | — | — | — | — | — | — | |
| FALCONBackbone=MagFace, Demographic-aware=false2026.06 | 94.1 | — | — | — | — | — | — | — | — | — | |
| GSTBackbone=MagFace, Demographic-aware=true2026.06 | 94.1 | — | — | — | — | — | — | — | — | — | |
| DemoNormBackbone=MagFace, Demographic-aware=true2026.06 | 94.1 | — | — | — | — | — | — | — | — | — | |
| OracleBackbone=MagFace, Demographic-aware=true2026.06 | 94.1 | — | — | — | — | — | — | — | — | — | |
| BaselineBackbone=ArcFace R50, demographic_labels_at_inference=false2026.06 | 94.08 | 95.63 | 48.81 | 66.2 | 24.4 | 17.13 | — | — | 53 | 1.7 | |
| OracleBackbone=ArcFace R50, demographic_labels_at_inference=true2026.06 | 94.08 | 95.85 | 51.04 | 68.42 | 2.17 | 7.59 | — | — | 47 | 1.2 | |
| GSTBackbone=ArcFace R50, demographic_labels_at_inference=true2026.06 | 94.08 | 95.7 | 51.14 | 68.43 | 2.31 | 7.7 | — | — | 47 | 1.23 | |
| DemoNormBackbone=ArcFace R50, demographic_labels_at_inference=true2026.06 | 94.08 | 95.74 | 50.7 | 68.56 | 15.44 | 14.37 | — | — | 47 | 1.17 | |
| FRAPPÉBackbone=MagFace, Demographic-aware=false2026.06 | 93.9 | — | — | — | — | — | — | — | — | — | |
| FRAPPEBackbone=ArcFace R50, demographic_labels_at_inference=false2026.06 | 93.86 | 95.7 | 51.55 | 66.4 | 2.16 | 7.71 | 3 | — | 43 | 1.4 | |
| FSNBackbone=MagFace, Demographic-aware=false2026.06 | 93.3 | — | — | — | — | — | — | — | — | — | |
| FSNBackbone=ArcFace R50, demographic_labels_at_inference=false2026.06 | 93.28 | 95.31 | 49.56 | 65.73 | 23.9 | 15.97 | 1 | — | 40 | 2.07 | |
| AC-RidgeBackbone=FaceNet, Dem-aware (require group labels at inference)=false2026.06 | 89.69 | 92.45 | 25.27 | 44.36 | 5.33 | 11.26 | 3 | — | — | — | |
| AC-MLPBackbone=FaceNet, Dem-aware (require group labels at inference)=false2026.06 | 88.13 | 90.81 | 17.52 | 36.73 | 3.32 | 12.31 | 2 | — | — | — | |
| FairCalBackbone=FaceNet, Dem-aware (require group labels at inference)=false2026.06 | 87.81 | 90.57 | 19.52 | 39.73 | 2.33 | 12.36 | 3 | — | — | — | |
| AC-LinearBackbone=FaceNet, Dem-aware (require group labels at inference)=false2026.06 | 87.8 | 90.61 | 16.98 | 37.6 | 3.48 | 12.41 | 1 | — | — | — | |
| FALCONBackbone=FaceNet, Dem-aware (require group labels at inference)=false2026.06 | 86.69 | 88.84 | 17.92 | 34.18 | 19.11 | 20 | 3 | — | — | — | |
| FSNBackbone=FaceNet, Dem-aware (require group labels at inference)=false2026.06 | 86.51 | 89.57 | 19.58 | 36.95 | 15.96 | 16.2 | 3 | — | — | — | |
| FRAPPEBackbone=FaceNet, Dem-aware (require group labels at inference)=false2026.06 | 86.39 | 89.57 | 19.07 | 37.4 | 2.07 | 13.03 | — | — | — | — | |
| BaselineBackbone=FaceNet, Dem-aware (require group labels at inference)=false2026.06 | 86.14 | 87.83 | 14.99 | 30.12 | 12.13 | 16.07 | — | — | — | — | |
| OracleBackbone=FaceNet, Dem-aware (require group labels at inference)=true2026.06 | 86.14 | 89.56 | 19.56 | 37.8 | 2.38 | 13.04 | — | — | — | — | |
| GSTBackbone=FaceNet, Dem-aware (require group labels at inference)=true2026.06 | 86.14 | 89.39 | 18.23 | 37.72 | 4.05 | 13.32 | — | — | — | — | |
| DemoNormBackbone=FaceNet, Dem-aware (require group labels at inference)=true2026.06 | 86.14 | 89.23 | 16.6 | 37.21 | 16.85 | 18.08 | — | — | — | — | |
| AC-DensityBackbone=R100, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 57 | — | — | |
| AC-DensityBackbone=AdaFace, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 65.4 | — | — | |
| AC-DensityBackbone=MagFace, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 39 | — | — | |
| AC-LinearBackbone=R100, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 59.4 | — | — | |
| AC-LinearBackbone=AdaFace, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 66.9 | — | — | |
| AC-LinearBackbone=MagFace, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 46.7 | — | — | |
| AC-MLPBackbone=R100, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 56.8 | — | — | |
| AC-MLPBackbone=AdaFace, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 64.7 | — | — | |
| AC-MLPBackbone=MagFace, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 42.5 | — | — | |
| BaselineBackbone=R100, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 56.9 | — | — | |
| BaselineBackbone=AdaFace, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 66.3 | — | — | |
| BaselineBackbone=MagFace, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 44.5 | — | — | |
| DemoNormBackbone=R100, Demographic Awareness=True2026.06 | — | — | — | — | — | — | — | 52 | — | — | |
| DemoNormBackbone=AdaFace, Demographic Awareness=True2026.06 | — | — | — | — | — | — | — | 65.2 | — | — | |
| DemoNormBackbone=MagFace, Demographic Awareness=True2026.06 | — | — | — | — | — | — | — | 42.7 | — | — | |
| FairCalBackbone=R100, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 57.5 | — | — | |
| FairCalBackbone=AdaFace, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 66 | — | — | |
| FairCalBackbone=MagFace, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 45.6 | — | — | |
| FALCONBackbone=R100, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 57.8 | — | — | |
| FALCONBackbone=AdaFace, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 66.6 | — | — | |
| FALCONBackbone=MagFace, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 45.9 | — | — | |
| FRAPPEBackbone=R100, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 57.5 | — | — | |
| FRAPPEBackbone=AdaFace, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 68.1 | — | — | |
| FRAPPEBackbone=MagFace, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 47.4 | — | — | |
| FSNBackbone=R100, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 53.6 | — | — | |
| FSNBackbone=AdaFace, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 62.3 | — | — | |
| FSNBackbone=MagFace, Demographic Awareness=False2026.06 | — | — | — | — | — | — | — | 43.9 | — | — | |
| GSTBackbone=R100, Demographic Awareness=True2026.06 | — | — | — | — | — | — | — | 54.6 | — | — | |
| GSTBackbone=AdaFace, Demographic Awareness=True2026.06 | — | — | — | — | — | — | — | 67.3 | — | — | |
| GSTBackbone=MagFace, Demographic Awareness=True2026.06 | — | — | — | — | — | — | — | 45.1 | — | — | |
| OracleBackbone=R100, Demographic Awareness=True2026.06 | — | — | — | — | — | — | — | 55.3 | — | — | |
| OracleBackbone=AdaFace, Demographic Awareness=True2026.06 | — | — | — | — | — | — | — | 67.1 | — | — | |
| OracleBackbone=MagFace, Demographic Awareness=True2026.06 | — | — | — | — | — | — | — | 44 | — | — |