Face Verification on YTF
98.06AccuracyBioMetricNet
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| BioMetricNet#Image=3.8M2018.01 | 98.06 | — | — | — | — | |
| BroadFaceTraining Set=IBUG500K, Backbone=ResNet100, #Image=11.96M2018.01 | 98.03 | — | — | — | — | |
| ArcFaceTraining Data=5.8M2019.04 | 98.02 | — | — | — | — | |
| ArcFaceTraining Set=MS1MV3, Backbone=ResNet100, #Image=5.1M2018.01 | 98.02 | — | — | — | — | |
| ArcFaceTraining Set=IBUG500K, Backbone=ResNet100, #Image=11.96M2018.01 | 98.01 | — | — | — | — | |
| BroadFace#Image=5.8M2018.01 | 98 | — | — | — | — | |
| GroupFace#Image=5.8M2018.01 | 97.8 | — | — | — | — | |
| upper-boundarysource=reproduced by authors2021.05 | 97.76 | — | — | — | — | |
| Partial-FCTraining Data=1% of MS1M identities2021.05 | 97.76 | — | — | — | — | |
| F2CTraining Data=1% of MS1M identities2021.05 | 97.76 | — | — | — | — | |
| DCQTraining Data=1% of MS1M identities2021.05 | 97.71 | — | — | — | — | |
| UniformFace#Image=6.1M2018.01 | 97.7 | — | — | — | — | |
| CosFaceTraining Data=5M, #Models=12018.01 | 97.6 | — | — | — | — | |
| CosFaceTraining Data=5M2019.04 | 97.6 | — | — | — | — | |
| CosFace#Image=5M2018.01 | 97.6 | — | — | — | — | |
| PFEfuse+matchTraining Data=4.4M2019.04 | 97.36 | — | — | — | — | |
| Noise-Tolerant#Image=1M noisy2018.01 | 97.36 | — | — | — | — | |
| PFE#Image=4.4M2018.01 | 97.36 | — | — | — | — | |
| PFEfuseTraining Data=4.4M2019.04 | 97.32 | — | — | — | — | |
| DeepFRTraining Data=2.6M, #Models=12018.01 | 97.3 | — | — | — | — | |
| VGG Face#Image=2.6M2018.01 | 97.3 | — | — | — | — | |
| BaselineTraining Data=4.4M2019.04 | 97.18 | — | — | — | — | |
| AFRN#Image=3.1M2018.01 | 97.1 | — | — | — | — | |
| RDCFace#Image=1.7M2018.01 | 97.1 | — | — | — | — | |
| DUL#Image=3.6M2018.01 | 96.78 | — | — | — | — | |
| RegularFace#Image=3.1M2018.01 | 96.7 | — | — | — | — | |
| CircleLoss#Image=3.6M2018.01 | 96.38 | — | — | — | — | |
| Fair Loss#Image=0.5M2018.01 | 96.2 | — | — | — | — | |
| L2-FaceTraining Data=3.7M2019.04 | 96.08 | — | — | — | — | |
| SFaceBackbone=ResNet50, Training Dataset=CASIA-WebFace, a=0.93, b=1.22022.05 | 96.08 | — | — | — | — | |
| ArcFaceBackbone=ResNet50, Training Dataset=CASIA-WebFace, m=0.42022.05 | 96.06 | — | — | — | — | |
| Unequal-training#Image=0.55M2018.01 | 96.04 | — | — | — | — | |
| SFaceBackbone=ResNet50, Training Dataset=CASIA-WebFace, a=0.9, b=1.32022.05 | 96 | — | — | — | — | |
| Marginal Loss#Image=3.8M2018.01 | 95.98 | — | — | — | — | |
| CombinedBackbone=ResNet50, Training Dataset=CASIA-WebFace, m=0.9,0.4,0.152022.05 | 95.86 | — | — | — | — | |
| SFaceBackbone=ResNet50, Training Dataset=CASIA-WebFace, a=0.9, b=1.22022.05 | 95.86 | — | — | — | — | |
| SFaceBackbone=ResNet50, Training Dataset=CASIA-WebFace, a=0.87, b=1.22022.05 | 95.82 | — | — | — | — | |
| ArcFaceBackbone=ResNet50, Training Dataset=CASIA-WebFace, m=0.32022.05 | 95.8 | — | — | — | — | |
| CosFaceBackbone=ResNet50, Training Dataset=CASIA-WebFace, m=0.352022.05 | 95.76 | — | — | — | — | |
| ArcFaceBackbone=ResNet50, Training Dataset=CASIA-WebFace, m=0.52022.05 | 95.66 | — | — | — | — | |
| softmaxBackbone=ResNet50, Training Dataset=CASIA-WebFace2022.05 | 95.6 | — | — | — | — | |
| NSoftmaxBackbone=ResNet50, Training Dataset=CASIA-WebFace, s=202022.05 | 95.54 | — | — | — | — | |
| D-softmaxBackbone=ResNet50, Training Dataset=CASIA-WebFace, d=0.92022.05 | 95.42 | — | — | — | — | |
| FaceNetTraining Data=200M, #Models=12018.01 | 95.1 | — | — | — | — | |
| FaceNetTraining Data=200M2019.04 | 95.1 | — | — | — | — | |
| FaceNetImages=200M2019.11 | 95.1 | — | — | — | — | |
| FaceNet#Image=200M2018.01 | 95.1 | — | — | — | — | |
| VFCTraining Data=1% of MS1M identities2021.05 | 95.08 | — | — | — | — | |
| SphereFaceTraining Data=0.49M, #Models=12018.01 | 95 | — | — | — | — | |
| SphereFaceTraining Data=0.5M2019.04 | 95 | — | — | — | — | |
| SphereFaceImages=0.7M2019.11 | 95 | — | — | — | — | |
| Single-task (pretrained)Images=1.1M2019.11 | 95 | — | — | — | — | |
| SphereFace#Image=0.5M2018.01 | 95 | — | — | — | — | |
| Center FaceTraining Data=0.7M, #Models=12018.01 | 94.9 | — | — | — | — | |
| CenterFaceTraining Data=0.7M2019.04 | 94.9 | — | — | — | — | |
| CenterlossImages=0.7M2019.11 | 94.9 | — | — | — | — | |
| Center Loss#Image=0.7M2018.01 | 94.9 | — | — | — | — | |
| Proposed dynamic MTLImages=1.1M2019.11 | 94.3 | — | — | — | — | |
| Single-task (fine-tuning)Images=1.1M2019.11 | 94.2 | — | — | — | — | |
| Static MTLImages=1.1M2019.11 | 94.1 | — | — | — | — | |
| Naive dynamic MTLImages=1.1M2019.11 | 94.1 | — | — | — | — | |
| TCP2021.05 | 93.92 | — | — | — | — | |
| lower-boundarysource=excerpted from VFC paper2021.05 | 93.78 | — | — | — | — | |
| Range Loss#Image=5M2018.01 | 93.7 | — | — | — | — | |
| DeepID2+Training Data=300K, #Models=252018.01 | 93.2 | — | — | — | — | |
| DeepID2+Training Data=300K2019.04 | 93.2 | — | — | — | — | |
| DeepID-2,3Images=-2019.11 | 93.2 | — | — | — | — | |
| DeepID#Image=0.2M2018.01 | 93.2 | — | — | — | — | |
| SphereFaceBackbone=ResNet50, Training Dataset=CASIA-WebFace, m=1.352022.05 | 93.18 | — | — | — | — | |
| Multi-similarity2021.05 | 92.76 | — | — | — | — | |
| N-pair2021.05 | 92.62 | — | — | — | — | |
| VGGFaceImages=2.6M2019.11 | 91.6 | — | — | — | — | |
| DeepFaceTraining Data=4M, #Models=32018.01 | 91.4 | — | — | — | — | |
| DeepFace+Training Data=4M2019.04 | 91.4 | — | — | — | — | |
| DeepFaceImages=4M2019.11 | 91.4 | — | — | — | — | |
| Deep Face#Image=4.4M2018.01 | 91.4 | — | — | — | — | |
| DeepFace-ensemble2016.12 | — | 91.4 | 96.3 | 92 | 54 | |
| MBGS LBP2016.12 | — | 74.7 | 82.6 | 60.5 | 35.8 |