3D Hand Mesh Reconstruction on HO3D v2 (test)
0.956F@5I2UV-HandNet
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| I2UV-HandNet2026.03 | 0.956 | 9.9 | 9.5 | 0.803 | — | — | — | — | — | |
| Hamba2026.03 | 0.648 | 7.5 | 7.7 | 0.982 | — | — | — | — | — | |
| WiLoR2026.03 | 0.646 | 7.5 | 7.7 | 0.983 | — | — | — | — | — | |
| HaMeR2023.12 | 0.635 | — | — | 0.98 | 0.846 | — | 0.841 | 7.7 | 7.9 | |
| HaMeR2026.03 | 0.635 | 7.7 | 7.9 | 0.98 | — | — | — | — | — | |
| Ours(final)Training Scheme=Supervised, Category=Probabilistic2023.04 | 0.608 | — | 8.2 | 0.965 | 0.835 | 8.3 | 0.836 | — | — | |
| AMVUR2023.12 | 0.608 | — | — | 0.965 | 0.835 | — | 0.836 | 8.3 | 8.2 | |
| Fast-HaMeRBackbone=ConvNeXt-L, Distillation Strategy=feature-level distillation, λKD=0.8, γFD=122026.03 | 0.599 | 8.1 | 8.3 | 0.979 | — | — | — | — | — | |
| HandOccNetTraining Scheme=Supervised, Category=Model-based2023.04 | 0.564 | — | 8.8 | 0.963 | 0.819 | 9.1 | 0.819 | — | — | |
| HandOccNet2023.12 | 0.564 | — | — | 0.963 | 0.819 | — | 0.819 | 9.1 | 8.8 | |
| HandOccNet2026.03 | 0.564 | 9.1 | 8.8 | 0.968 | — | — | — | — | — | |
| MobReconBackbone=DenseStack2021.12 | 0.538 | 9.2 | 9.4 | 0.957 | — | — | — | — | — | |
| Chen et al.Training Scheme=Supervised, Category=Model-free2023.04 | 0.538 | — | 9.4 | 0.957 | — | 9.2 | — | — | — | |
| MobRecon2023.12 | 0.538 | — | — | 0.957 | — | — | — | 9.2 | 9.4 | |
| Ours(AMVUR)Training Scheme=Supervised, Category=Model-free2023.04 | 0.533 | — | 9.4 | 0.958 | 0.814 | 9.3 | 0.813 | — | — | |
| Liu et al.Training Scheme=Supervised, Category=Model-based2023.04 | 0.528 | — | 9.5 | 0.956 | 0.803 | 9.9 | 0.81 | — | — | |
| Liu et al.2023.12 | 0.528 | — | — | 0.956 | 0.803 | — | 0.81 | 9.9 | 9.5 | |
| HO3DBackbone=CPM2021.12 | 0.506 | 10.7 | 10.6 | 0.942 | — | — | — | — | — | |
| Hampali et al.Training Scheme=Supervised, Category=Model-based2023.04 | 0.506 | — | 10.6 | 0.942 | 0.788 | 10.7 | 0.79 | — | — | |
| Hampali et al.2023.12 | 0.506 | — | — | 0.942 | 0.788 | — | 0.79 | 10.7 | 10.6 | |
| I2UV-HandNetBackbone=ResNet502021.12 | 0.5 | 9.9 | 10.1 | 0.943 | — | — | — | — | — | |
| I2UV-HandNetTraining Scheme=Supervised, Category=Model-based2023.04 | 0.5 | — | 10.1 | 0.943 | 0.804 | 9.9 | 0.799 | — | — | |
| I2UV-HandNet2023.12 | 0.5 | — | — | 0.943 | 0.804 | — | 0.799 | 9.9 | 10.1 | |
| MobReconBackbone=GhostStack2021.12 | 0.488 | 10 | 10.2 | 0.948 | — | — | — | — | — | |
| ArtiBoostTraining Scheme=Supervised, Category=Model-based2023.04 | 0.488 | — | 10.9 | 0.944 | 0.773 | 11.4 | 0.782 | — | — | |
| ArtiBoost2023.12 | 0.488 | — | — | 0.944 | 0.773 | — | 0.782 | 11.4 | 10.9 | |
| ArtiBoost2026.03 | 0.488 | 11.1 | 10.9 | 0.944 | — | — | — | — | — | |
| METROTraining Scheme=Supervised, Category=Model-free2023.04 | 0.484 | — | 11.1 | 0.946 | 0.792 | 10.4 | 0.779 | — | — | |
| METRO2023.12 | 0.484 | — | — | 0.946 | 0.792 | — | 0.779 | 10.4 | 11.1 | |
| METRO2026.03 | 0.484 | 10.4 | 11.1 | 0.946 | — | — | — | — | — | |
| Ours(final)Training Scheme=Weakly-Supervised, Category=Probabilistic2023.04 | 0.48 | — | 10.8 | 0.949 | 0.787 | 10.3 | 0.784 | — | — | |
| ObManBackbone=ResNet182021.12 | 0.464 | 11 | 11 | 0.939 | — | — | — | — | — | |
| Hasson et al.Training Scheme=Supervised, Category=Model-based2023.04 | 0.464 | — | 11.2 | 0.939 | 0.78 | 11 | 0.777 | — | — | |
| Hasson et al.2023.12 | 0.464 | — | — | 0.939 | 0.78 | — | 0.777 | 11 | 11.2 | |
| Ours(prior-net)Training Scheme=Supervised, Category=Model-based2023.04 | 0.46 | — | 11.5 | 0.936 | 0.783 | 10.9 | 0.77 | — | — | |
| Pose2MeshTraining Scheme=Supervised, Category=Model-free2023.04 | 0.441 | — | 12.7 | 0.909 | 0.754 | 12.5 | 0.749 | — | — | |
| Pose2Mesh2023.12 | 0.441 | — | — | 0.909 | 0.754 | — | 0.749 | 12.5 | 12.7 | |
| Pose2Mesh2026.03 | 0.441 | 12.5 | 12.7 | 0.909 | — | — | — | — | — | |
| S2HANDTraining Scheme=Weakly-Supervised, Category=Model-based2023.04 | 0.44 | — | — | 0.93 | 0.765 | — | 0.769 | — | — | |
| Ours(prior-net)Training Scheme=Weakly-Supervised, Category=Model-based2023.04 | 0.417 | — | 12 | 0.925 | 0.752 | 12.4 | 0.76 | — | — | |
| I2L-MeshNetTraining Scheme=Supervised, Category=Model-free2023.04 | 0.409 | — | 13.9 | 0.932 | 0.775 | 11.2 | 0.722 | — | — | |
| I2L-MeshNet2023.12 | 0.409 | — | — | 0.932 | 0.775 | — | 0.722 | 11.2 | 13.9 | |
| I2L-MeshNet2026.03 | 0.409 | 11.2 | 13.9 | 0.932 | — | — | — | — | — | |
| Ours(AMVUR)Training Scheme=Weakly-Supervised, Category=Model-free2023.04 | 0.375 | — | 15.1 | 0.907 | 0.778 | 10.8 | 0.698 | — | — | |
| Keypoint TransTraining Scheme=Supervised, Category=Model-free2023.04 | — | — | — | — | 0.786 | 10.8 | — | — | — | |
| Keypoint Trans2023.12 | — | — | — | — | 0.786 | — | — | 10.8 | — | |
| THOR-Net2026.03 | — | 11.3 | 10.7 | — | — | — | — | — | — |