Research map / 3D vision / Human pose estimation
Neural Voting Field for Camera-Space 3D Hand Pose Estimation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use unified 3D dense point-wise voting instead of holistic regression | Camera-space hand pose estimation · FreiHAND | CS-MJE: 54.5 -> 47.2 | Helped |
| Use unified 3D dense point-wise voting instead of 2D dense pixel-wise voting | Camera-space hand pose estimation · FreiHAND | CS-MJE: 53.2 -> 47.2 | Helped |
| Use hand scale (ablation) | CS-NVF · FreiHAND | CS-MJE: 47.2 -> 42.4 | Helped |
| Pre-train on Comp (ablation) | CS-NVF · FreiHAND | CS-MJE: 47.2 -> 44.6 | Helped |
| Use both hand scale and Comp pre-training (ablation) | CS-NVF · FreiHAND | CS-MJE: 42.4 -> 39.3 | Helped |
| Use 50% of near-surface points with highest voting weight (ablation) | CS-NVF · FreiHAND | CS-MJE: 40.2 -> 39.3 | Helped |
| Increase voting fraction from 50% to 75% (ablation) | CS-NVF · FreiHAND | CS-MJE: 39.3 -> 39.5 | Hurt |
| Increase clamping distance from 5 mm to 10 mm (ablation) | CS-NVF · FreiHAND | CS-MJE: 39.3 -> 40.0 | Hurt |
| Increase KNN from 1024 to 2048 (ablation) | CS-NVF · FreiHAND | CS-MJE: 39.3 -> 39.3 | No clear effect |
| Increase sampling step size from 16 mm to 32 mm (ablation) | CS-NVF · FreiHAND | CS-MJE: 39.3 -> 42.4 | Hurt |
| Increase sampling step size from 16 mm to 32 mm (ablation) | CS-NVF · FreiHAND | FPS: 34 -> 56 | Helped |
| Use NVF for root-relative pose estimation rather than the strongest listed baseline | Root-relative hand pose estimation · HO3D | MJE: 24.0 -> 21.8 | Helped |
Closest papers by meaning
- Dense 3D Regression for Hand Pose Estimation (CVPR 2018 · similarity 0.93)
- Pushing the Envelope for RGB-Based Dense 3D Hand Pose Estimation via Neural Rendering (CVPR 2019 · similarity 0.92)
- Depth-Based 3D Hand Pose Estimation: From Current Achievements to Future Goals (CVPR 2018 · similarity 0.91)
- SRHand: Super-Resolving Hand Images and 3D Shapes via View/Pose-aware Neural Image Representations and Explicit Meshes (NeurIPS 2025 · similarity 0.89)
- Keypoint Fusion for RGB-D Based 3D Hand Pose Estimation (AAAI 2024 · similarity 0.89)
- HOISDF: Constraining 3D Hand-Object Pose Estimation with Global Signed Distance Fields (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.