List your company
ResearchFormerly WizWand

Facial Landmark Detection on 300-W public (test)

2.62NMEGV(HRNet)
2.39083.93795.4857.0321Mar 18, 2018Oct 26, 2018Jun 6, 2019Jan 15, 2020Aug 24, 2020Apr 4, 2021Nov 13, 2021
Updated 8mo ago

Evaluation results

MethodLinks
GV(HRNet)backbone=HRNet2021.112.62
GEAN2021.112.68
3DDE2021.112.69
LRefNets2021.112.71
AWing2021.112.72
DCFE2021.112.76
HRNetV2021.112.87
LAB2021.112.98
PFLD 1Xscale=1X2021.113.01
ResNet50-FFLDbackbone=ResNet-502021.113.06
DAN2021.113.19
CCNN2021.113.23
SAN2021.113.34
efnbackbone=EfficientNet-B32021.113.34
CPM2021.113.39
ODN2021.113.56
mnv2KDbackbone=MobileNetV22021.113.56
PCD-CNN2021.113.67
mnv2backbone=MobileNetV22021.113.93
DSRN2021.114.12
RCN2021.114.67
RARDL=true, number of parameters=15.65M+, model memory=62.6MB+2018.034.94
SIRDL=true, number of parameters=3.72M, model memory=15.6MB2018.035.04
MDMDL=true, number of parameters=80.56M, model memory=322.3MB2018.035.61
CFSSDL=false, model memory=225.2MB2018.035.76
baseline CRDL=true, number of parameters=14.88M, model memory=62.4MB2018.036.23
LBFDL=false, model memory=36.6MB2018.036.32
RCPRDL=false, model memory=91.3MB2018.038.35