List your company
ResearchFormerly WizWand

Facial Landmark Detection on 300-W Common Subset

2.46Mean Error (%)SHT-M-i
2.15444.21726.288.3428Aug 18, 2014Jul 13, 2016Jun 8, 2018May 4, 2020Mar 30, 2022Feb 23, 2024Jan 19, 2026
Updated 8mo ago

Evaluation results

MethodLinks
SHT-M-iInput Resolution=64x64, Extra Training Data=CelebA2026.012.46
SHT-M-vInput Resolution=64x64, Extra Training Data=300VW2026.012.5
STAR+2026.012.52
ADNet2026.012.53
SHT-MInput Resolution=64x642026.012.57
DTLD+2026.012.6
AWing2026.012.72
GlomFce2026.012.72
DHLN-MInput Resolution=64x642026.012.74
SLPT2026.012.75
LUVLi2026.012.76
HG-HSLE2026.012.85
HRNet2020.062.87
HR-Net2026.012.87
Structure Coherence Component2020.062.88
DU-Net2026.012.9
Techer2026.012.91
DeCaFA2020.062.93
DeCaFa2026.012.93
LAB2020.062.98
LAB2026.012.98
SHNInput Resolution=256x2562026.013.11
AVS2026.013.21
CPM+SBR2020.063.28
SAN2020.063.34
SAN2026.013.34
PCD-CNN2020.063.67
CFSSlandmarks=682014.084.73
TCDCNlandmarks=682014.084.8
LBFlandmarks=682014.084.95
ESRlandmarks=682014.085.28
CFANlandmarks=682014.085.5
SDMlandmarks=682014.085.57
GN-DPMlandmarks=682014.085.78
RCPRlandmarks=682014.086.18
DRMFlandmarks=682014.086.65
CDMlandmarks=682014.0810.1