Face Verification on CFP (Frontal-Frontal)
99.886AccuracyProdpoly-ResNet50
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Prodpoly-ResNet50Model=Prodpoly-ResNet502020.06 | 99.886 | — | — | |
| ResNet50Model=ResNet502020.06 | 99.871 | — | — | |
| GroupFaceBackbone=ResNet101, Training Dataset=MS1MV22021.05 | 98.63 | — | — | |
| PartialFC-r0.1Backbone=ResNet101, Training Dataset=MS1MV2, Sampling Rate=0.12021.05 | 98.6 | — | — | |
| PartialFC-r1.0Backbone=ResNet101, Training Dataset=MS1MV2, Sampling Rate=1.02021.05 | 98.51 | — | — | |
| VarGFaceNetParams.(M)=5.0, MFLOPS=10222021.08 | 98.5 | — | — | |
| MagFaceBackbone=ResNet-100, Resolution=112x1122021.03 | 98.46 | — | — | |
| DCQBackbone=ResNet101, Training Dataset=MS1MV22021.05 | 98.44 | — | — | |
| ArcFaceBackbone=ResNet-100, Resolution=112x1122021.03 | 98.4 | — | — | |
| CosFace (ours)Backbone=ResNet101, Training Dataset=MS1MV22021.05 | 98.38 | — | — | |
| ArcfaceTraining epoch=16, Params=65M, MAdds=12.1G, Batch size=5122019.08 | 98.37 | — | — | |
| CurricularFaceBackbone=ResNet1002020.04 | 98.37 | — | — | |
| CurricularFaceBackbone=ResNet101, Training Dataset=MS1MV22021.05 | 98.37 | — | — | |
| MV-Arc-Softmax*Backbone=ResNet1002020.04 | 98.28 | — | — | |
| ArcFace2020.04 | 98.27 | — | — | |
| ArcFaceBackbone=ResNet101, Training Dataset=MS1MV22021.05 | 98.27 | — | — | |
| CosFaceBackbone=ResNet-100, Resolution=112x1122021.03 | 98.26 | — | — | |
| SoftmaxBackbone=ResNet-100, Resolution=112x1122021.03 | 98.2 | — | — | |
| VarGFaceNetFlops=1022M2019.10 | 98.086 | — | — | |
| y2Flops=933M2019.10 | 97.829 | — | — | |
| ShuffleFaceNet 2xParams.(M)=4.5, MFLOPS=10502021.08 | 97.56 | — | — | |
| DW-SeesawFaceNet V2Training epoch=16, Params=4.2M, MAdds=526M, Batch size=1282019.08 | 97.24 | — | — | |
| CNN2016.04 | 96.93 | 2.51 | 99.68 | |
| ShuffleFaceNet 1.5xParams.(M)=2.6, MFLOPS=577.52021.08 | 96.9 | — | — | |
| MobileFaceNetParams.(M)=2.0, MFLOPS=9332021.08 | 96.9 | — | — | |
| SphereFaceBackbone=ResNet-100, Resolution=112x1122021.03 | 96.84 | — | — | |
| Sengupta et al.2016.04 | 96.4 | 3.48 | 99.43 | |
| DW-SeesawFaceNet V1Training epoch=16, Params=4.1M, MAdds=514M, Batch size=1282019.08 | 96.39 | — | — | |
| Human Accuracy2016.04 | 96.24 | 5.34 | 98.19 | |
| MobileFaceNetsParams.(M)=0.99, MFLOPS=439.82021.08 | 96.07 | — | — | |
| MobileFaceNetV1Params.(M)=3.4, MFLOPS=11002021.08 | 95.8 | — | — | |
| PocketNetM-256Params.(M)=1.75, MFLOPS=1099.152021.08 | 95.66 | — | — | |
| ShrinkTeaNet-MFNRParams=3.73M, Batch size=5122019.08 | 95.14 | — | — | |
| SV-AM-SoftmaxBackbone=ResNet-100, Resolution=112x1122021.03 | 95.1 | — | — | |
| PocketNetM-128Params.(M)=1.68, MFLOPS=1099.022021.08 | 95.07 | — | — | |
| ProxylessFaceNASParams.(M)=3.2, MFLOPS=9002021.08 | 94.7 | — | — | |
| AirFaceTraining epoch=61, MAdds=1G, Batch size=10242019.08 | 94.514 | — | — | |
| PocketNetS-128Params.(M)=0.92, MFLOPS=587.112021.08 | 94.21 | — | — | |
| Distill-DSE-LSEParams.(M)=1.352021.08 | 94.19 | — | — | |
| Deng et al.2020.04 | 94.05 | — | — | |
| Dream2020.04 | 93.98 | — | — | |
| Peng et al.2020.04 | 93.76 | — | — | |
| PocketNetS-256Params.(M)=0.99, MFLOPS=587.242021.08 | 93.34 | — | — | |
| Seesaw-shareFaceNetBatch size=160, Training epoch=16, Params=1.3M, MAdds=146M2019.08 | 93.09 | — | — | |
| Seesaw-shuffleFaceNet(mobi)Batch size=224, Training epoch=16, Params=2.8M, MAdds=154M2019.08 | 93.07 | — | — | |
| Seesaw-shuffleFaceNetBatch size=192, Training epoch=16, Params=1.3M, MAdds=146M2019.08 | 92.9 | — | — | |
| ShuffleFaceNet 0.5xParams.(M)=0.5, MFLOPS=66.92021.08 | 92.59 | — | — | |
| Seesaw-shuffleFaceNetBatch size=160, Training epoch=16, Params=1.3M, MAdds=146M2019.08 | 92.03 | — | — | |
| MobileFaceNet (ours)Batch size=256, Training epoch=24, Params=1.2M, MAdds=221M2019.08 | 91.81 | — | — | |
| MobileFaceNet (ours)Batch size=160, Training epoch=16, Params=1.2M, MAdds=221M2019.08 | 91.64 | — | — | |
| MobileFaceNet (ours)Batch size=256, Training epoch=16, Params=1.2M, MAdds=221M2019.08 | 91.46 | — | — | |
| MobileFaceNet (ours)Batch size=192, Training epoch=16, Params=1.2M, MAdds=221M2019.08 | 91.39 | — | — | |
| DRGAN2020.04 | 90.11 | — | — | |
| LMobileNetE#Image=3.8M, Training epoch=27, Params=26.7M, Batch size=5122019.08 | 88.5 | — | — |