Face Verification on AgeDB-30
9,765AccuracyViT
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ViTBackbone=ViT-B, Train data=WebFace4M2026.06 | 9,765 | — | |
| ViT-8RBackbone=ViT-B + 8 Registers, Train data=WebFace4M2026.06 | 9,740 | — | |
| TopoFRTraining Data=Glint360K, Backbone=R200, Venue=NeurIPS242024.10 | 98.82 | — | |
| TopoFRTraining Data=Glint360K, Backbone=R100, Venue=NeurIPS242024.10 | 98.72 | — | |
| TransFace-LTraining Data=Glint360K, Backbone=ViT, Venue=ICCV232024.10 | 98.62 | — | |
| AdaFaceTraining Data=Glint360K, Backbone=R200, Venue=CVPR222024.10 | 98.61 | — | |
| VPL-ArcFaceBackbone=ResNet101, Train data=MS1MV32026.06 | 98.6 | — | |
| AdaFaceTraining Data=Glint360K, Backbone=R100, Venue=CVPR222024.10 | 98.58 | — | |
| TopoFR†Training Data=MS1MV2, Backbone=R200, Venue=NeurIPS242024.10 | 98.54 | — | |
| TransFace-BTraining Data=Glint360K, Backbone=ViT, Venue=ICCV232024.10 | 98.53 | — | |
| TransFaceBackbone=TransFace-B, Train data=MS1MV22026.06 | 98.53 | — | |
| TopoFRTraining Data=MS1MV2, Backbone=R200, Venue=NeurIPS242024.10 | 98.52 | — | |
| ArcFaceTraining Data=Glint360K, Backbone=R200, Venue=CVPR192024.10 | 98.49 | — | |
| TopoFRTraining Data=Glint360K, Backbone=R50, Venue=NeurIPS242024.10 | 98.47 | — | |
| Prodpoly-ResNet50Model=Prodpoly-ResNet502020.06 | 98.467 | — | |
| TopoFRTraining Data=MS1MV2, Backbone=R100, Venue=NeurIPS242024.10 | 98.42 | — | |
| TopoFR†Training Data=MS1MV2, Backbone=R100, Venue=NeurIPS242024.10 | 98.42 | — | |
| BroadFaceTraining Data=MS1MV2, Backbone=R100, Venue=ECCV202024.10 | 98.38 | — | |
| BroadFaceBackbone=ResNet101, Train data=MS1MV22026.06 | 98.38 | — | |
| ElasticFace-ArcTraining Dataset=MS1MV2 [7], [4]2021.09 | 98.35 | — | |
| ElasticFace-Arc+Training Dataset=MS1MV2 [7], [4]2021.09 | 98.35 | — | |
| ViT-8RBackbone=ViT-B + 8 Registers, Train data=MS1MV32026.06 | 98.35 | — | |
| AdaFaceTraining Data=Glint360K, Backbone=R50, Venue=CVPR222024.10 | 98.34 | — | |
| TransFace++-BBackbone=ViT-B, Train data=MS1MV22026.06 | 98.34 | — | |
| ResNet100Bits=FP322022.06 | 98.33 | — | |
| CurricularFaceTraining Dataset=MS1MV2 [7], [4]2021.09 | 98.32 | — | |
| CurricularFaceTraining Data=MS1MV2, Backbone=R100, Venue=CVPR202024.10 | 98.32 | — | |
| CurricularFaceBackbone=ResNet101, Train data=MS1MV22026.06 | 98.32 | — | |
| ArcFaceTraining Data=Glint360K, Backbone=R100, Venue=CVPR192024.10 | 98.31 | — | |
| SCF-ArcFaceTraining Data=MS1MV2, Backbone=R100, Venue=CVPR212024.10 | 98.3 | — | |
| Part-based FRBackbone=Part fViT-B, Train data=MS1MV32026.06 | 98.29 | — | |
| GroupFaceTraining Dataset=clean MS1M [7], [4]2021.09 | 98.28 | — | |
| ElasticFace-Cos+Training Dataset=MS1MV2 [7], [4]2021.09 | 98.28 | — | |
| ResNet-100 ElasticFace (Cos+)Category=Baseline, Params [M]=65.22023.07 | 98.28 | — | |
| ElasticFace-Cos+Training Data=MS1MV2, Backbone=R100, Venue=CVPR222024.10 | 98.28 | — | |
| AdaFaceTraining Data=MS1MV2, Backbone=R200, Venue=CVPR222024.10 | 98.28 | — | |
| ArcFaceTraining Data=Glint360K, Backbone=R50, Venue=CVPR192024.10 | 98.28 | — | |
| ArcFaceBackbone=ResNet101, Train data=MS1MV22026.06 | 98.28 | — | |
| ElasticFace-Cos+Backbone=ResNet101, Train data=MS1MV22026.06 | 98.28 | — | |
| ElasticFace-CosTraining Dataset=MS1MV2 [7], [4]2021.09 | 98.27 | — | |
| ElasticFace (MS1MV2)Augmentation=false, Identities=85 K, Total training images=5,800 K2023.08 | 98.27 | — | |
| TransFace-BTraining Data=MS1MV2, Backbone=ViT, Venue=ICCV232024.10 | 98.27 | — | |
| TopoFRTraining Data=MS1MV2, Backbone=R50, Venue=NeurIPS242024.10 | 98.25 | — | |
| ViTBackbone=ViT-B, Train data=MS1MV32026.06 | 98.25 | — | |
| ResNet50Model=ResNet502020.06 | 98.233 | — | |
| TopoFR†Training Data=MS1MV2, Backbone=R50, Venue=NeurIPS242024.10 | 98.23 | — | |
| TransFace-LTraining Data=MS1MV2, Backbone=ViT, Venue=ICCV232024.10 | 98.23 | — | |
| Partial-FC-ArcFaceTraining Dataset=MS1MV2 [7], [4]2021.09 | 98.2 | — | |
| ArcFaceTraining Data=MS1MV2, Backbone=R200, Venue=CVPR192024.10 | 98.19 | — | |
| CosFaceBackbone=ResNet-100, Resolution=112x1122021.03 | 98.17 | — | |
| MagFaceBackbone=ResNet-100, Resolution=112x1122021.03 | 98.17 | — | |
| MagFaceTraining Dataset=MS1MV2 [7], [4]2021.09 | 98.17 | — | |
| CosFaceTraining Data=MS1MV2, Backbone=R100, Venue=CVPR182024.10 | 98.17 | — | |
| MagFace+Training Data=MS1MV2, Backbone=R100, Venue=CVPR212024.10 | 98.17 | — | |
| MagFaceBackbone=ResNet101, Train data=MS1MV22026.06 | 98.17 | — | |
| AdaFaceBackbone=ResNet101, Train data=MS1MV32026.06 | 98.17 | — | |
| ArcfaceTraining epoch=16, Params=65M, MAdds=12.1G, Batch size=5122019.08 | 98.15 | — | |
| ArcFace (LResNet100E-IR)FLOPs (M)=24211, Parameters (M)=65.22021.07 | 98.15 | — | |
| VarGFaceNetParams.(M)=5.0, MFLOPS=10222021.08 | 98.15 | — | |
| ArcFaceTraining Dataset=MS1MV2 [7], [4]2021.09 | 98.15 | — | |
| ResNet-100 ArcFaceCategory=Baseline, Params [M]=65.22023.07 | 98.15 | — | |
| VarGFaceNetCategory=Baseline, Params [M]=5.02023.07 | 98.15 | — | |
| ArcFaceTraining Data=MS1MV2, Backbone=R100, Venue=CVPR192024.10 | 98.15 | — | |
| SwinFaceBackbone=Swin-T, Train data=MS1MV22026.06 | 98.15 | — | |
| ResNet100Quantization data=Real data, Bits=w8a82022.06 | 98.13 | — | |
| CosFaceBackbone=ResNet101, Train data=MS1MV22026.06 | 98.11 | — | |
| VarGFaceNetFlops=1022M2019.10 | 98.1 | — | |
| VarGFaceNetFLOPs (M)=1022, Parameters (M)=52021.07 | 98.1 | — | |
| ResNet50Bits=FP322022.06 | 98.08 | — | |
| ArcFaceBackbone=ResNet-100, Resolution=112x1122021.03 | 98.05 | — | |
| AdaFaceTraining Data=MS1MV2, Backbone=R100, Venue=CVPR222024.10 | 98.05 | — | |
| ViT-FREE12Exit Depth=12, Fine-tuning Protocol=None, GFLOPs=11.502026.06 | 98.05 | — | |
| AdaFaceBackbone=ResNet101, Train data=MS1MV22026.06 | 98.05 | — | |
| Face TransformerBackbone=ViT-P12S8, Train data=MS1MV22026.06 | 98.05 | — | |
| Partial-FC-CosFaceTraining Dataset=MS1MV2 [7], [4]2021.09 | 98.03 | — | |
| ResNet50Quantization data=Real data, Bits=w8a82022.06 | 98 | — | |
| ViT-8RBackbone=ViT-B + 8 Registers, Train data=MS1MV22026.06 | 98 | — | |
| KP-RPEBackbone=ViT-B+KP-RPE, Train data=MS1MV32026.06 | 97.98 | — | |
| QuantFaceBackbone=ResNet50, Quantization data=Synthetic data, Bits=w8a82022.06 | 97.97 | — | |
| QuantFaceBackbone=ResNet100, Quantization data=Synthetic data, Bits=w8a82022.06 | 97.95 | — | |
| MV-SoftmaxTraining Data=MS1MV2, Backbone=R100, Venue=AAAI202024.10 | 97.95 | — | |
| ArcFaceBackbone=ResNet101, Train data=WebFace4M2026.06 | 97.95 | — | |
| ViT-FREE11Exit Depth=11, Fine-tuning Protocol=None, GFLOPs=10.552026.06 | 97.92 | — | |
| AdaFaceBackbone=ResNet101, Train data=WebFace4M2026.06 | 97.9 | — | |
| AdaFaceTraining Data=MS1MV2, Backbone=R50, Venue=CVPR222024.10 | 97.85 | — | |
| ViTBackbone=ViT-B, Train data=MS1MV22026.06 | 97.78 | — | |
| Dyn-arcFaceTraining Dataset=clean MS1M [7], [4]2021.09 | 97.76 | — | |
| SoftmaxBackbone=ResNet-100, Resolution=112x1122021.03 | 97.72 | — | |
| MagFaceTraining Data=MS1MV2, Backbone=R50, Venue=CVPR212024.10 | 97.7 | — | |
| KP-RPEBackbone=ViT-B+KP-RPE, Train data=WebFace4M2026.06 | 97.67 | — | |
| MobileFaceNetFLOPs (M)=933, Parameters (M)=22021.07 | 97.6 | — | |
| MobileFaceNetParams.(M)=2.0, MFLOPS=9332021.08 | 97.6 | — | |
| MobileFaceNetCategory=Baseline, Params [M]=2.02023.07 | 97.6 | — | |
| DW-SeesawFaceNet V2Training epoch=16, Params=4.2M, MAdds=526M, Batch size=1282019.08 | 97.55 | — | |
| ArcFaceTraining Data=MS1MV2, Backbone=R50, Venue=CVPR192024.10 | 97.53 | — | |
| DW-SeesawFaceNet V1Training epoch=16, Params=4.1M, MAdds=514M, Batch size=1282019.08 | 97.52 | — | |
| y2Flops=933M2019.10 | 97.517 | — | |
| KP-RPE-8R*Backbone=ViT-B+KP-RPE + 8 Registers, Train data=WebFace4M2026.06 | 97.47 | — | |
| QuantFaceBackbone=ResNet50, Quantization data=Synthetic data, Bits=w6a62022.06 | 97.43 | — | |
| KP-RPE*Backbone=ViT-B+KP-RPE, Train data=WebFace4M2026.06 | 97.35 | — |