Face Recognition on IJB-C
97.85TAR (FAR=1e-4)TransFace-L
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TransFace-LYear='25, Param (M)=271.6, FLOPs=25.4G, Training Dataset=Glint360K, Model Scale=Large-Scale, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=OOM2026.04 | 97.85 | — | |
| TransFace-BYear='25, Param (M)=124.5, FLOPs=11.5G, Training Dataset=Glint360K, Model Scale=Large-Scale, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=18.202026.04 | 97.59 | — | |
| TransFace-SYear='25, Param (M)=86.7, FLOPs=5.8G, Training Dataset=Glint360K, Model Scale=Large-Scale, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=14.312026.04 | 97.45 | — | |
| ResNet50-ArcFaceYear='22, Param (M)=43.6, FLOPs=6.3G, Training Dataset=Glint360K, Model Scale=Large-Scale, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=3.762026.04 | 96.81 | — | |
| FaceLiVTv2-LYear=-, Param (M)=8.52, FLOPs=309M, Training Dataset=Glint360K, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=0.71(14.0× ↓)2026.04 | 96.59 | — | |
| ResNet200-TopoFRYear='24, Param (M)=118.8, FLOPs=23.5G, Training Dataset=Glint360K, Model Scale=Large-Scale, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=12.242026.04 | 96.56 | — | |
| FaceLiVTv2-MYear=-, Param (M)=7.02, FLOPs=258M, Training Dataset=Glint360K, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=0.65(16.7%↓)2026.04 | 96.42 | — | |
| FaceLiVTv2-SYear=-, Param (M)=4.62, FLOPs=179M, Training Dataset=Glint360K, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=0.54(30.8%↓)2026.04 | 95.99 | — | |
| FaceLiVTv1-MYear='25, Param (M)=9.8, FLOPs=386M, Training Dataset=Glint360K, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=0.672026.04 | 95.7 | — | |
| KANFace-0.5Year='25, Param (M)=6.80, FLOPs=397M, Training Dataset=WebFace12M, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=9.982026.04 | 95.64 | — | |
| EdgeFace-SYear='24, Param (M)=3.6, FLOPs=306M, Training Dataset=WebFace12M, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=9.892026.04 | 95.63 | — | |
| ArcFaceVenue=CVPR '19, PPFR=No2024.03 | 95.6 | — | |
| DuetFaceVenue=MM '22, PPFR=Yes2024.03 | 95.3 | — | |
| GhostFaceNetV1-1Year='23, Param (M)=4.1, FLOPs=216M, Training Dataset=MS1MV3, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=0.782026.04 | 94.94 | — | |
| PartialFaceVenue=ICCV '23, PPFR=Yes2024.03 | 94.93 | — | |
| LOTR-MFace detector=PyramidKey [74]2021.09 | 94.92 | — | |
| LOTR-R+Face detector=PyramidKey [74]2021.09 | 94.89 | — | |
| LOTR-M+Face detector=PyramidKey [74]2021.09 | 94.88 | — | |
| EdgeFace-XSYear='24, Param (M)=1.77, FLOPs=154M, Training Dataset=WebFace12M, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=5.822026.04 | 94.85 | — | |
| LOTR-MFace detector=RetinaFace [67]2021.09 | 94.79 | — | |
| LOTR-M+Face detector=RetinaFace [67]2021.09 | 94.77 | — | |
| KANFace-0.6Year='25, Param (M)=4.74, FLOPs=240M, Training Dataset=WebFace12M, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=6.522026.04 | 94.75 | — | |
| LOTR-R+Face detector=RetinaFace [67]2021.09 | 94.74 | — | |
| MinusFaceVenue=ours, PPFR=Yes, Backbone=IR-502024.03 | 94.7 | — | |
| SwiftFaceFormer-L3Year='24, Param (M)=28.0, FLOPs=2.0G, Training Dataset=MS1MV3, Model Scale=Large-Scale, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=2.592026.04 | 94.7 | — | |
| VarGFaceNetYear='19, Param (M)=5.0, FLOPs=1.0G, Training Dataset=MS1MV3, Model Scale=Large-Scale, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=0.832026.04 | 94.7 | — | |
| MobileFaceNetYear='21, Param (M)=0.99, FLOPs=440M, Training Dataset=MS1MV2, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=0.772026.04 | 94.7 | — | |
| EdgeFace-SYear='24, Param (M)=3.6, FLOPs=306M, Training Dataset=Glint360K, Model Scale=Lightweight, Experimental Setup=Retrained from scratch, Latency (ms) on iPhone 15 Pro=9.892026.04 | 94.66 | — | |
| PyramidKeyFace landmarks=From detector2021.09 | 94.59 | — | |
| RetinaFaceFace landmarks=From detector2021.09 | 94.59 | — | |
| DCTDPVenue=ECCV '22, PPFR=Yes, Privacy Budget=epsilon=12024.03 | 94.43 | — | |
| KANFace-0.5Year='25, Param (M)=6.80, FLOPs=397M, Training Dataset=Glint360K, Model Scale=Lightweight, Experimental Setup=Retrained from scratch, Latency (ms) on iPhone 15 Pro=9.982026.04 | 94.34 | — | |
| ShuffleFaceNet-1.5Year='21, Param (M)=2.6, FLOPs=577M, Training Dataset=MS1MV2, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=0.692026.04 | 94.3 | — | |
| PPFR-FDVenue=AAAI '22, PPFR=Yes2024.03 | 94.07 | — | |
| SwiftFaceFormer-L1Year='24, Param (M)=11.8, FLOPs=805M, Training Dataset=MS1MV3, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=1.202026.04 | 93.82 | — | |
| SwiftFaceFormer-SYear='24, Param (M)=6.0, FLOPs=485M, Training Dataset=MS1MV3, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=0.652026.04 | 93.54 | — | |
| GhostFaceNetV1-2Year='23, Param (M)=4.1, FLOPs=60M, Training Dataset=MS1MV3, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=0.602026.04 | 93.45 | — | |
| MixFaceNets-MYear='21, Param (M)=3.9, FLOPs=626M, Training Dataset=MS1MV2, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=0.702026.04 | 93.42 | — | |
| GhostFaceNetV2-2Year='23, Param (M)=6.8, FLOPs=77M, Training Dataset=MS1MV3, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=0.672026.04 | 93.16 | — | |
| PocketNetM256Year='22, Param (M)=1.75, FLOPs=1.1G, Training Dataset=CASIA-WF, Model Scale=Large-Scale, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=0.982026.04 | 92.7 | — | |
| FaceLiVTv1-SYear='25, Param (M)=5.89, FLOPs=237M, Training Dataset=Glint360K, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=0.472026.04 | 92.7 | — | |
| PocketNetM128Year='22, Param (M)=1.68, FLOPs=1.1G, Training Dataset=CASIA-WF, Model Scale=Large-Scale, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=0.982026.04 | 92.63 | — | |
| SwiftFaceFormer-XSYear='24, Param (M)=3.4, FLOPs=294M, Training Dataset=MS1MV3, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=0.592026.04 | 92.32 | — | |
| PocketNetS128Year='22, Param (M)=0.92, FLOPs=587M, Training Dataset=CASIA-WF, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=0.882026.04 | 91.62 | — | |
| PocketNetS256Year='22, Param (M)=0.99, FLOPs=587M, Training Dataset=CASIA-WF, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=0.882026.04 | 91.33 | — | |
| FaceLiVTv2-XSYear=-, Param (M)=2.9, FLOPs=90M, Training Dataset=Glint360K, Model Scale=Lightweight, Experimental Setup=Original, Latency (ms) on iPhone 15 Pro=0.43(35.8%↓)2026.04 | 91.25 | — | |
| AdvFaceVenue=CVPR '23, PPFR=Yes2024.03 | 74.39 | — | |
| ProFaceVenue=MM '22, PPFR=Yes2024.03 | 72.96 | — | |
| InstaHideVenue=ICML '20, PPFR=Yes2024.03 | 69.02 | — | |
| CloakVenue=WWW '21, PPFR=Yes, Trade-off parameter=1002024.03 | 33.82 | — | |
| PEEPVenue=CS '20, PPFR=Yes, Privacy Budget=52024.03 | 6.02 | — | |
| SapiensIDTraining Data=WB4M, Aligned=true2025.04 | — | 96.43 | |
| ViT-AdaFaceTraining Data=WebFace4M, Aligned=true2025.04 | — | 97.14 | |
| ViT-AdaFaceTraining Data=WB4M (Face crop), Aligned=true2025.04 | — | 97.22 |