List your company
ResearchFormerly WizWand

Facial Landmark Detection on MAFL (test)

2.42Normalised MSE (%)Deep Latent Particles
1.88325.50669.1312.7534Jun 20, 2018Feb 15, 2019Oct 13, 2019Jun 9, 2020Feb 4, 2021Oct 2, 2021May 31, 2022
Updated 8mo ago

Evaluation results

MethodLinks
Deep Latent ParticlesK=50, features=mean + log-variance2022.052.42
Deep Latent ParticlesK=50, features=mean2022.052.43
Deep Latent ParticlesK=30, features=mean + log-variance2022.052.49
Deep Latent ParticlesK=25, features=mean + log-variance2022.052.52
Proposed MethodTraining Set=CelebA, Loss Network=self-supervised, K=502018.062.54
KeyNetK=502022.052.54
Deep Latent ParticlesK=30, features=mean2022.052.56
Proposed MethodTraining Set=CelebA, Loss Network=self-supervised, K=302018.062.58
KeyNetK=302022.052.58
Proposed MethodTraining Set=CelebA, Loss Network=supervised, K=502018.062.59
Proposed MethodTraining Set=CelebA, Loss Network=supervised, K=302018.062.63
Deep Latent ParticlesK=25, features=mean2022.052.87
ZhangSupervision=Unsupervised, K=30, Equivariance=true2018.063.16
ZhangK=302022.053.16
Proposed MethodTraining Set=CelebA, Loss Network=self-supervised, K=102018.063.19
Proposed MethodTraining Set=CelebA, Loss Network=supervised, K=102018.063.32
WilesSupervision=Unsupervised, Dataset=VoxCeleb2, Regressor Bias=true2018.063.44
ZhangSupervision=Unsupervised, K=102018.063.46
Proposed MethodTraining Set=VoxCeleb, Loss Network=self-supervised, K=30, Regressor Bias=true2018.063.63
Proposed MethodTraining Set=VoxCeleb, Loss Network=self-supervised, K=302018.063.94
Proposed MethodTraining Set=VoxCeleb, Loss Network=supervised, K=302018.064.01
MTCNNSupervision=Supervised2018.065.39
Shu +Supervision=Unsupervised2018.065.45
Thewlis (frames)Supervision=Unsupervised2018.065.83
ThewlisSupervision=Unsupervised, K=502018.066.67
ThewlisSupervision=Unsupervised, K=302018.067.15
TCDCNSupervision=Supervised2018.067.95
ZhangSupervision=Unsupervised, K=30, Equivariance=false2018.068.42
Cascaded CNNSupervision=Supervised2018.069.73
CFANSupervision=Supervised2018.0615.84