Details for PyMAF-X SMPL-X

AuthorsHongwen Zhang, Yating Tian, Yuxiang Zhang, Mengcheng Li, Liang An, Zhenan Sun, Yebin Liu
Paper titlePyMAF-X: Towards Well-aligned Full-body Model Regression from Monocular Images
Publication venueIEEE TPAMI
URLhttps://www.liuyebin.com/pymaf-x
Additional Information an eXpressive version of PyMAF [ICCV'21, https://arxiv.org/abs/2103.16507 ] for monocular human/hand/face/full-body mesh recovery

Metrics

Full setKid
F1 Score0.890F1 Score0.420
Precision0.900Precision0.270
Recall0.890Recall0.920
Body
Full setKid
MPJPE83.200MPJPE145.200
MVE84.000MVE146.700
NMJE93.500NMJE345.700
NMVE94.400NMVE349.300
Face
Full setKid
MPJPE37.900MPJPE54.900
MVE35.000MVE52.400
Full body
Full setKid
MPJPE124.600MPJPE194.900
MVE125.700MVE196.500
NMJE140.000NMJE219.000
NMVE141.200NMVE220.800
Hands
Full setKid
left hand MPJPE42.500left hand MPJPE46.200
left hand MVE44.600left hand MVE47.600
right hand MPJPE43.700right hand MPJPE47.900
right hand MVE45.600right hand MVE49.300