Human Mesh Reconstruction on Human3.6M
28.5PA-MPJPEMQ-HMR
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| MQ-HMRVenue=Ours2025.01 | 28.5 | 42.5 | — | — | |
| Virtual MarkersIntermediate Representation=Virtual marker, Temporal cues=false2023.03 | 32 | 47.3 | 58 | — | |
| CLIFFTemporal cues=false2023.03 | 32.7 | 47.1 | — | — | |
| CLIFFVenue=ECCV 20222025.01 | 32.7 | 47.1 | — | — | |
| PointHMRBackbone=HRNet-W32, Model Category=model-free2023.04 | 32.9 | 48.3 | — | — | |
| HMR2.0Venue=ICCV 20232025.01 | 33.6 | 44.8 | — | — | |
| FastMETROBackbone=HRNet-W64, Model Category=model-free2023.04 | 33.7 | 52.2 | — | — | |
| FastMETROIntermediate Representation=3D vertices, Temporal cues=false2023.03 | 33.7 | 52.2 | — | — | |
| FastMETROVenue=ECCV 20222025.01 | 33.7 | 52.2 | — | — | |
| HybrIKBackbone=ResNet-34, Model Category=model-based2023.04 | 34.5 | 54.4 | — | — | |
| MeshGraphormerBackbone=HRNet-W64, Model Category=model-free2023.04 | 34.5 | 51.2 | — | — | |
| HybrIKIntermediate Representation=3D skeleton, Temporal cues=false2023.03 | 34.5 | 54.4 | 65.7 | — | |
| Mesh GraphormerIntermediate Representation=3D vertices, Temporal cues=false2023.03 | 34.5 | 51.2 | — | — | |
| VisDBIntermediate Representation=3D vertices, Temporal cues=false2023.03 | 34.5 | 51 | — | — | |
| Mesh Graphormer2021.04 | 34.5 | 51.2 | — | — | |
| TokenHMRVenue=CVPR 20242025.01 | 36.3 | 48.4 | — | — | |
| METROBackbone=HRNet-W64, Model Category=model-free2023.04 | 36.7 | 54 | — | — | |
| METROIntermediate Representation=3D vertices, Temporal cues=false2023.03 | 36.7 | 54 | — | — | |
| METRO2021.04 | 36.7 | 54 | — | — | |
| MAEDBackbone=ResNet-50, Model Category=model-based2023.04 | 38.7 | 56.4 | — | — | |
| MAEDIntermediate Representation=3D skeleton, Temporal cues=true2023.03 | 38.7 | 56.3 | — | — | |
| DecoMRIntermediate Representation=IUV image, Temporal cues=false2023.03 | 39.3 | 60.6 | — | — | |
| THUNDRIntermediate Representation=3D markers, Temporal cues=false2023.03 | 39.8 | 55 | — | — | |
| PyMAFBackbone=ResNet-50, Model Category=model-based2023.04 | 40.5 | 57.7 | — | — | |
| PyMafIntermediate Representation=IUV image, Temporal cues=false2023.03 | 40.5 | 57.7 | — | — | |
| SPINBackbone=ResNet-50, Model Category=model-based2023.04 | 41.1 | — | — | — | |
| TCMRTemporal cues=true2023.03 | 41.1 | 62.3 | — | — | |
| SPINTemporal cues=false2023.03 | 41.1 | — | — | — | |
| I2L-MeshNetIntermediate Representation=3D vertices, Temporal cues=false2023.03 | 41.1 | 55.7 | 65.1 | — | |
| SPIN2021.04 | 41.1 | — | — | — | |
| I2LMeshNet2021.04 | 41.1 | 55.7 | — | — | |
| ProHMRTemporal cues=false2023.03 | 41.2 | — | — | — | |
| VIBEBackbone=ResNet-50, Model Category=model-based2023.04 | 41.4 | 65.6 | — | — | |
| VIBE2021.04 | 41.4 | 65.6 | — | — | |
| VIBETemporal cues=true2023.03 | 41.5 | 65.9 | — | — | |
| I2LMeshNetBackbone=ResNet-50, Model Category=model-free2023.04 | 41.7 | 55.7 | — | — | |
| DSD-SATNIntermediate Representation=3D skeleton, Temporal cues=true2023.03 | 42.4 | 59.1 | — | — | |
| Pose2MeshIntermediate Representation=3D skeleton, Temporal cues=false2023.03 | 46.3 | 64.9 | 85.3 | — | |
| Pose2MeshBackbone=HRNet-W48, Model Category=model-free2023.04 | 47 | 64.9 | — | — | |
| Pose2Mesh2021.04 | 47 | 64.9 | — | — | |
| DenseRacIntermediate Representation=IUV image, Temporal cues=false2023.03 | 48 | 76.8 | — | — | |
| GraphCMRBackbone=ResNet-50, Model Category=model-free2023.04 | 50.1 | — | — | — | |
| GraphCMRIntermediate Representation=3D vertices, Temporal cues=false2023.03 | 50.1 | — | — | — | |
| GraphCMR2021.04 | 50.1 | — | — | — | |
| PAREVenue=ICCV 20212025.01 | 50.6 | 76.8 | — | — | |
| Arnab et al.Intermediate Representation=2D skeleton, Temporal cues=true2023.03 | 54.3 | 77.8 | — | — | |
| HMRBackbone=ResNet-50, Model Category=model-based2023.04 | 56.8 | 88 | — | — | |
| HMRTemporal cues=false2023.03 | 56.8 | 88 | 96.1 | — | |
| HMR2021.04 | 56.8 | 88 | — | — | |
| HMMRTemporal cues=true2023.03 | 56.9 | — | — | — | |
| LMTBackbone=ResNet-50, Resolution=224 x 2242022.08 | — | 30.56 | 42.28 | 14.61 | |
| LMTBackbone=ResNet-152, Resolution=384 x 3842022.08 | — | 17.59 | 23.7 | 11.33 | |
| LT-fittingBackbone=ResNet-152, Resolution=384 x 3842022.08 | — | 16.21 | 35.2 | 15.73 | |
| Parameter regr.Backbone=ResNet-50, Resolution=224 x 2242022.08 | — | 46.9 | — | — |