Research map / 3D vision / 3D generation
Crowd3D: Towards Hundreds of People Reconstruction From a Single Image
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 Crowd3D instead of BEV-Large | Large-scene crowd reconstruction · LargeCrowd | PPDS: 74.21 -> 81.53 | Helped |
| Use Crowd3D instead of BEV-Large | Large-scene crowd reconstruction · LargeCrowd | PA-PPDS: 75.05 -> 89.36 | Helped |
| Use Crowd3D instead of BEV-Large | Large-scene crowd reconstruction · LargeCrowd | PCOD: 87.31 -> 92.63 | Helped |
| Use Crowd3D instead of BEV-Large | Large-scene crowd reconstruction · LargeCrowd | OKS: 66.15 -> 71.72 | Helped |
| Add HVIP-based ground transform instead of ankle-joint midpoint transform (ablation) | Crowd3D · LargeCrowd | PPDS: 80.45 -> 81.53 | Helped |
| Add HVIP-based ground transform instead of ankle-joint midpoint transform (ablation) | Crowd3D · LargeCrowd | PA-PPDS: 88.95 -> 89.36 | Helped |
| Add HVIP-based ground transform instead of ankle-joint midpoint transform (ablation) | Crowd3D · LargeCrowd | PCOD: 92.42 -> 92.63 | Helped |
| Add HVIP-based ground transform instead of ankle-joint midpoint transform (ablation) | Crowd3D · LargeCrowd | OKS: 64.17 -> 71.72 | Helped |
| Use adaptive human-centric cropping instead of uniform cropping (ablation) | Crowd3D · LargeCrowd | cropping score: 0.806 -> 0.923 | Helped |
| Estimate ground and camera parameters using more than ten people (ablation) | Crowd3D camera and ground-plane estimation · LargeCrowd | ground-normal cosine distance | No clear effect |
| Vary the number of people used for calibration (ablation) | Crowd3D camera and ground-plane estimation · LargeCrowd | focal length RMSE | No clear effect |
Closest papers by meaning
- Crowd4D: Scene-Aware Monocular 4D Crowd Reconstruction (ICML 2026 · similarity 0.89)
- CrowdGaussian: Reconstructing High-Fidelity 3D Gaussians for Human Crowd from a Single Image (CVPR 2026 · similarity 0.88)
- Reconstructing People, Places, and Cameras (CVPR 2025 · similarity 0.86)
- Guess The Unseen: Dynamic 3D Scene Reconstruction from Partial 2D Glimpses (CVPR 2024 · similarity 0.85)
- ODHSR: Online Dense 3D Reconstruction of Humans and Scenes from Monocular Videos (CVPR 2025 · similarity 0.84)
- MultiPly: Reconstruction of Multiple People from Monocular Video in the Wild (CVPR 2024 · similarity 0.84)
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.