Research map / Point clouds / Shape and surface completion
HORT: Monocular Hand-held Objects Reconstruction with Transformers
Zerui Chen, Rolandos Alexandros Potamias, Shizhe Chen, Cordelia Schmid
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 hand encoder with 3D hand vertices in palm frame (ablation) | HORT · ObMan (30K train) | FS@5: 0.45 -> 0.53 | Improved |
| transform hand vertices into joint coordinate systems (ablation) | HORT · ObMan (30K train) | FS@5: 0.53 -> 0.60 | Improved |
| fine-tune DINOv2 instead of training ViT from scratch (ablation) | HORT · ObMan (30K train) | CD: 2.6 -> 1.8 | Improved |
| fine-tune DINOv2 instead of freezing it (ablation) | HORT · ObMan (30K train) | CD: 2.9 -> 1.8 | Improved |
| jointly decode sparse points and hand-relative translation (ablation) | HORT · ObMan (30K train) | CD: 3.8 -> 3.0 | Improved |
| add dense upsampling decoder (ablation) | HORT · ObMan (30K train) | CD: 3.0 -> 2.3 | Improved |
| replace global image feature with pixel-aligned features (ablation) | HORT · ObMan (30K train) | CD: 2.3 -> 2.0 | Improved |
| add hand features to dense decoder (ablation) | HORT · ObMan (30K train) | CD: 2.0 -> 1.8 | Improved |
| increase hand pose noise sigma 0.0 to 0.5 (ablation) | HORT · DexYCB | CD: 1.0 -> 1.6 | Hurt |
| convert point cloud to mesh | HORT · DexYCB | FS@10: 0.85 | No clear effect |
| HORT vs D-SCO on HO3D | HORT · HO3D | CD: 3.4 -> 2.6 | Improved |
| HORT vs D-SCO inference speed | HORT · ObMan | fps: 0.08 -> 16.9 | Improved |
Closest papers by meaning
- Learning Explicit Contact for Implicit Reconstruction of Hand-Held Objects from Monocular Images (AAAI 2024 · similarity 0.89)
- Reconstructing Hands in 3D with Transformers (CVPR 2024 · similarity 0.88)
- HOLD: Category-agnostic 3D Reconstruction of Interacting Hands and Objects from Video (CVPR 2024 · similarity 0.87)
- In-Hand 3D Object Reconstruction from a Monocular RGB Video (AAAI 2024 · similarity 0.87)
- HOISDF: Constraining 3D Hand-Object Pose Estimation with Global Signed Distance Fields (CVPR 2024 · similarity 0.87)
- SPAR3D: Stable Point-Aware Reconstruction of 3D Objects from Single Images (CVPR 2025 · similarity 0.87)
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.
