Face Recognition on SLFW
99.81AccuracyFC-R100
Evaluation results
| Method | Links | |
|---|---|---|
| FC-R100Training Dataset=Webface42M, Backbone=ResNet-1002021.05 | 99.81 | |
| FC-R50Training Dataset=Glint360k, Backbone=ResNet-502021.05 | 99.71 | |
| FC-R50Training Dataset=MS1MV2, Backbone=ResNet-502021.05 | 99.55 | |
| upper-boundarysource=reproduced by authors2021.05 | 99.55 | |
| F2C-R50Training Dataset=Glint360k, Backbone=ResNet-502021.05 | 99.53 | |
| F2C-R50Training Dataset=MS1MV2, Backbone=ResNet-502021.05 | 99.45 | |
| Partial-FCTraining Data=1% of MS1M identities2021.05 | 99.28 | |
| DCQTraining Data=1% of MS1M identities2021.05 | 99.23 | |
| F2CTraining Data=1% of MS1M identities2021.05 | 99.23 | |
| FC-MobileTraining Dataset=MS1MV2, Backbone=Mobile2021.05 | 98.8 | |
| F2C-R100Training Dataset=Webface42M, Backbone=ResNet-1002021.05 | 98.8 | |
| F2C-MobileTraining Dataset=MS1MV2, Backbone=Mobile2021.05 | 98.57 | |
| VFCTraining Data=1% of MS1M identities2021.05 | 96.23 | |
| lower-boundarysource=excerpted from VFC paper2021.05 | 93.52 | |
| TCP2021.05 | 93.23 | |
| N-pair2021.05 | 92.28 | |
| Multi-similarity2021.05 | 91.03 |