List your company
ResearchFormerly WizWand

Face Verification on CALFW (test)

96.21AccuracyMag+UNPG
90.344491.867293.3994.9128May 24, 2021Jul 13, 2021Sep 1, 2021Oct 22, 2021Dec 11, 2021Jan 30, 2022Mar 22, 2022
Updated 8mo ago

Evaluation results

MethodLinks
Mag+UNPGBackbone=ResNet-1002022.0396.21
GroupFaceBackbone=ResNet101, Training Dataset=MS1MV22021.0596.2
CurricularFaceBackbone=ResNet101, Training Dataset=MS1MV22021.0596.2
PartialFC-r1.0Backbone=ResNet101, Training Dataset=MS1MV2, Sampling Rate=1.02021.0596.2
CosFace (ours)Backbone=ResNet101, Training Dataset=MS1MV22021.0596.2
MagFace*Backbone=ResNet-1002022.0396.15
Cos+UNPGBackbone=ResNet-1002022.0396.15
MagFaceBackbone=ResNet-1002022.0396.15
Arc+UNPGBackbone=ResNet-1002022.0396.13
PartialFC-r0.1Backbone=ResNet101, Training Dataset=MS1MV2, Sampling Rate=0.12021.0596.12
CosFaceBackbone=ResNet-1002022.0396.11
ArcFaceBackbone=ResNet-1002022.0396.11
DCQBackbone=ResNet101, Training Dataset=MS1MV22021.0596.07
ArcFaceBackbone=ResNet101, Training Dataset=MS1MV22021.0595.45
ArcFace*Backbone=ResNet-1002022.0395.45
CosFaceBackbone=ResNet101, Training Dataset=MS1MV22021.0590.57