Face Verification on LFW unrestricted with labeled outside data 9
99.65AccuracyFaceNet
Evaluation results
| Method | Links | |
|---|---|---|
| FaceNetModels=1, Data=200M*2017.04 | 99.65 | |
| DeepID2+Models=25, Data=300K*2017.04 | 99.47 | |
| SphereFaceModels=1, Data=WebFace2017.04 | 99.42 | |
| Center FaceModels=1, Data=0.7M*2017.04 | 99.28 | |
| BaiduModels=1, Data=1.3M*2017.04 | 99.13 | |
| L-Softmax LossModels=1, Data=WebFace2017.04 | 99.1 | |
| Softmax+Center LossModels=1, Data=WebFace2017.04 | 99.05 | |
| Deep FRModels=1, Data=2.6M2017.04 | 98.95 | |
| Softmax+ContrastiveModels=1, Data=WebFace2017.04 | 98.78 | |
| Liu et al.Models=1, Data=WebFace2017.04 | 98.71 | |
| DeepID2+Models=1, Data=300K*2017.04 | 98.7 | |
| Triplet LossModels=1, Data=WebFace2017.04 | 98.7 | |
| Ding et al.Models=1, Data=WebFace2017.04 | 98.43 | |
| Softmax LossModels=1, Data=WebFace2017.04 | 97.88 | |
| Yi et al.Models=1, Data=WebFace2017.04 | 97.73 | |
| DeepFaceModels=3, Data=4M*2017.04 | 97.35 |