Research map / 3D vision / 3D generation
Reconstructing Humans and Objects in Interaction using Large Reconstruction Models
Agniv Chatterjee, Georgios Pavlakos
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 |
|---|---|---|---|
| use joint human-object LRM scaffold instead of independent reconstruction (ablation) | MILO · InterCap | PA-CD h+o: 15.89 -> 9.36 | Improved |
| use joint LRM scaffold instead of object-only EasyHOI-style fitting (ablation) | MILO · InterCap | PA-CD h+o: 37.21 -> 9.36 | Improved |
| add pose fitting after root fitting (ablation) | MILO · InterCap (no template) | PA-CD h+o: 10.15 -> 9.36 | Improved |
| add root fitting over raw LRM mesh (ablation) | MILO · InterCap (no template) | PA-CD h+o: 14.10 -> 10.15 | Improved |
| replace Hunyuan3D-2.0 with SAM3D (joint) (ablation) | MILO · InterCap | PA-CD h+o: 9.36 -> 9.80 | No clear effect |
| replace Hunyuan3D-2.0 with InstantMesh (ablation) | MILO · InterCap | PA-CD h+o: 9.36 -> 10.84 | Hurt |
| use oracle GT mesh instead of LRM mesh (ablation) | MILO · InterCap | PA-CD h+o: 9.36 -> 3.12 | Improved |
| add template alignment to LRM object | MILO · InterCap | PA-CD object: 20.74 -> 18.97 | Improved |
| add template alignment to LRM object | MILO · InterCap | PA-CD human: 6.85 -> 6.96 | No clear effect |
| add template alignment for large objects (chair, trolley, umbrella) | MILO · InterCap per-object | PA-CD object: chair 12.22 -> 18.77 | Hurt |
| add ICP after Kabsch in template alignment (ablation) | MILO · InterCap | PA-CD h+o: 7.53 -> 7.45 | Improved |
| use own segmentation instead of P3-SAM (ablation) | MILO · InterCap | PA-CD object: 55.86 -> 20.74 | Improved |
Closest papers by meaning
- Reconstructing In-the-Wild Open-Vocabulary Human-Object Interactions (CVPR 2025 · similarity 0.90)
- PICO: Reconstructing 3D People In Contact with Objects (CVPR 2025 · similarity 0.89)
- Reconstructing People, Places, and Cameras (CVPR 2025 · similarity 0.88)
- End-to-End HOI Reconstruction Transformer with Graph-based Encoding (CVPR 2025 · similarity 0.88)
- HOI-M^3: Capture Multiple Humans and Objects Interaction within Contextual Environment (CVPR 2024 · similarity 0.88)
- Real2Sim in HOI: Toward Physically Plausible HOI Reconstruction from Monocular Videos (NeurIPS 2026 · 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.
