Research map / 3D vision / Human pose estimation
HandVoxNet++: 3D Hand Shape and Pose Estimation Using Voxel-Based Neural Networks
Jameel Nawaz Malik, Soshi Shimada, Ahmed Elhayek, Sk Aziz Ali, Christian Theobalt, Vladislav Golyanik, Didier Stricker
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add joint heatmaps input to voxelized depth (ablation) | V2S-Net · SynHand5M | 3D vertex error (mm): 8.78 -> 3.36 | Improved |
| add voxelized depth input to joint heatmaps (ablation) | V2S-Net · SynHand5M | 3D vertex error (mm): 3.54 -> 3.36 | Improved |
| add joint heatmaps input (ablation) | V2V-ShapeNet · SynHand5M | 3D shape error: 0.007 -> 0.005 | Improved |
| add voxelized depth input (ablation) | V2V-ShapeNet · SynHand5M | 3D shape error: 0.016 -> 0.005 | Improved |
| replace DispVoxNet registration with GCN-MeshReg | HandVoxNet · SynHand5M | 3D vertex error (mm): 2.92 -> 1.72 | Improved |
| add refinement step to GCN-MeshReg (ablation) | HandVoxNet · SynHand5M | 3D vertex error (mm): 1.79 -> 1.72 | Improved |
| add GCN-MeshReg registration (ablation) | HandVoxNet++ · HANDS19 shape | 3D vertex error (mm): 4.90 -> 3.57 | Improved |
| add refinement step to GCN-MeshReg (ablation) | HandVoxNet++ · HANDS19 shape | 3D vertex error (mm): 4.08 -> 3.57 | Improved |
| use GCN-MeshReg instead of classical NRGA++ registration (ablation) | HandVoxNet++ · HANDS19 shape | 3D vertex error (mm): 4.5 -> 3.57 | Improved |
| use GCN-MeshReg instead of DispVoxNet registration | HandVoxNet · HANDS19 shape | 3D vertex error (mm): 4.14 -> 3.57 | Improved |
| use projective TSDF instead of projective D-TSDF (ablation) | HandVoxNet++ · HANDS19 pose | main error (mm): 13.94 -> 13.35 | Improved |
| use projective TSDF instead of occupancy grid | HandVoxNet++ vs HandVoxNet · HANDS19 pose | main error (mm): 15.57 -> 13.35 | Improved |
Later papers that compared with this method
2 later accepted papers report HandVoxNet++ as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- HandDAGT: A Denoising Adaptive Graph Transformer for 3D Hand Pose Estimation (ECCV 2024 · better on 1 of 1 shared result)
- Two Heads Are Better than One: Image-Point Cloud Network for Depth-Based 3D Hand Pose Estimation (AAAI 2023 · better on 1 of 1 shared result)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- HandVoxNet: Deep Voxel-Based Network for 3D Hand Shape and Pose Estimation From a Single Depth Map (CVPR 2020 · similarity 0.98)
- V2V-PoseNet: Voxel-to-Voxel Prediction Network for Accurate 3D Hand and Human Pose Estimation From a Single Depth Map (CVPR 2018 · similarity 0.93)
- 3D Hand Shape and Pose Estimation From a Single RGB Image (CVPR 2019 · similarity 0.91)
- 3D Hand Shape and Pose From Images in the Wild (CVPR 2019 · similarity 0.90)
- Real-Time 3D Hand Pose Estimation with 3D Convolutional Neural Networks (TPAMI 2018 · similarity 0.90)
- Reconstructing Hands in 3D with Transformers (CVPR 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
