Research map / 3D vision / Face and shape
Feed-Forward Multi-view Multi-person Reconstruction with Contrastive Human-Aware 3D Representation
Yuanwang Yang, Buzhen Huang, Zongxuan Ren, Jing Huang, Kun Li
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 unified 3D feature space (ablation) | full framework · EgoHumans | CA-MPJPE: 3.12 -> 0.95 | Improved |
| add unified 3D feature space (ablation) | full framework · EgoHumans | PA-MPJPE: 2.53 -> 0.18 | Improved |
| add reprojection-based enhancement (ablation) | full framework · EgoHumans | CA-MPJPE: 1.28 -> 0.95 | Improved |
| add reprojection-based enhancement (ablation) | full framework · EgoHumans | PA-MPJPE: 1.74 -> 0.18 | Improved |
| add spatial contrastive learning (SCL) (ablation) | full framework · EgoHumans | CA-MPJPE: 1.94 -> 0.95 | Improved |
| add spatial contrastive learning (SCL) (ablation) | full framework · EgoHumans | GA-MPJPE: 0.49 -> 0.30 | Improved |
| replace VGGT backbone with AnySplat (ablation) | full framework · EgoHumans | CA-MPJPE: 0.95 -> 1.02 | No clear effect |
| learned 3D human confidence instead of 2D mask fusion (ablation) | token-level human confidence · single-person dataset | Precision: 0.420 -> 0.886 | Improved |
| learned 3D human confidence instead of 2D mask fusion | token-level human confidence · single-person dataset | Recall: 0.989 -> 0.825 | Hurt |
| learned 3D human confidence instead of 2D mask fusion | token-level human confidence · single-person dataset | PR-AUC: 0.006 -> 0.863 | Improved |
| use spatial contrastive features instead of VGGT pose-aware ReID | association · cross-view association | AIDP: 82.02 -> 98.54 | Improved |
| feed-forward method instead of HSfM optimization | Ours vs HSfM · runtime, 1 input frame | Time (s): 170 -> 0.209 | Improved |
Closest papers by meaning
- Multiview Human Body Reconstruction from Uncalibrated Cameras (NeurIPS 2022 · similarity 0.90)
- Human Interaction-Aware 3D Reconstruction from a Single Image (CVPR 2026 · similarity 0.90)
- Instance-aware Contrastive Learning for Occluded Human Mesh Reconstruction (CVPR 2024 · similarity 0.90)
- Coherent Reconstruction of Multiple Humans From a Single Image (CVPR 2020 · similarity 0.89)
- MUC: Mixture of Uncalibrated Cameras for Robust 3D Human Body Reconstruction (AAAI 2025 · similarity 0.88)
- MultiPly: Reconstruction of Multiple People from Monocular Video in the Wild (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.
