HyperAI超神経

Multi Hypotheses 3D Human Pose Estimation On 2

評価指標

Best-Hypothesis MPJPE (n = 25)
Best-Hypothesis PMPJPE (n = 25)
H36M PMPJPE (n = 1)
H36M PMPJPE (n = 25)
Most-Likely Hypothesis PMPJPE (n = 1)

評価結果

このベンチマークにおける各モデルのパフォーマンス結果

モデル名
Best-Hypothesis MPJPE (n = 25)
Best-Hypothesis PMPJPE (n = 25)
H36M PMPJPE (n = 1)
H36M PMPJPE (n = 25)
Most-Likely Hypothesis PMPJPE (n = 1)
Paper TitleRepository
ProHMR-60.141.236.867.3Probabilistic Modeling for Human Mesh Recovery
SMPL-MDN (by 3D Multi-bodies)91.569.544.842.774.7Generating Multiple Hypotheses for 3D Human Pose Estimation with Mixture Density Network
ProHMR (2D Vis, by MHEntropy)-82.6-64.3-Probabilistic Modeling for Human Mesh Recovery
HuManiFlow-----HuManiFlow: Ancestor-Conditioned Normalising Flows on SO(3) Manifolds for Human Pose and Shape Distribution Estimation
3D Multi-bodies90.064.241.642.267.83D Multi-bodies: Fitting Sets of Plausible 3D Human Models to Ambiguous Image Data-
MHEntropy (3D)-50.6-36.8-MHEntropy: Entropy Meets Multiple Hypotheses for Pose and Shape Recovery-
SMPL-CVAE109.775.146.746.276.53D Multi-bodies: Fitting Sets of Plausible 3D Human Models to Ambiguous Image Data-
HMR--56.856.8-End-to-end Recovery of Human Shape and Pose
MHEntropy (2D Vis)-66.4-51.3-MHEntropy: Entropy Meets Multiple Hypotheses for Pose and Shape Recovery-
HMR (2D Vis, by MHEntropy)-85.267.467.485.2End-to-end Recovery of Human Shape and Pose
0 of 10 row(s) selected.