Research map / 3D vision / 3D generation
Pose2Room: Understanding 3D Scenes from Human Activities
Yinyu Nie, Angela Dai, Xiaoguang Han, Matthias Nießner
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 the relative position encoder to the baseline (ablation) | P2R-Net · S1 | mAP@0.5: 8.71 -> 34.43 | Improved |
| add the relative position encoder to the baseline (ablation) | P2R-Net · S2 | mAP@0.5: 3.12 -> 19.07 | Improved |
| add the spatio-temporal pose encoder (ablation) | P2R-Net · S1 | mAP@0.5: 34.43 -> 39.98 | Improved |
| add the spatio-temporal pose encoder (ablation) | P2R-Net · S2 | mAP@0.5: 19.07 -> 29.25 | Improved |
| add the probabilistic mixture decoder (ablation) | P2R-Net · S1 | mAP@0.5: 39.98 -> 42.20 | Improved |
| add the probabilistic mixture decoder (ablation) | P2R-Net · S2 | mAP@0.5: 29.25 -> 35.34 | Improved |
| use full mixture decoder instead of a single-mode Gaussian variant (-G) (ablation) | P2R-Net · S1 | TMD: 1.73 -> 3.34 | Improved |
| use full mixture decoder instead of a heatmap-style variant (-H) (ablation) | P2R-Net · S1 | mAP@0.5: 31.41 -> 42.20 | Improved |
| pretrain on the synthetic dataset before PROX | P2R-Net · PROX | mAP@0.5: 5.36 -> 31.38 | Improved |
| increase Gaussian input noise from sigma to 10 sigma | P2R-Net · S1 | mAP@0.5: 38.58 -> 31.71 | Hurt |
| increase Gaussian input noise from sigma to 10 sigma | P2R-Net · S2 | mAP@0.5: 27.16 -> 26.18 | No clear effect |
Closest papers by meaning
- Populating 3D Scenes by Learning Human-Scene Interaction (CVPR 2021 · similarity 0.86)
- Learning 3D Human Dynamics From Video (CVPR 2019 · similarity 0.86)
- Modeling 3D Environments through Hidden Human Context (TPAMI 2016 · similarity 0.86)
- Perceiving 3D Human-Object Spatial Arrangements from a Single Image in the Wild (ECCV 2020 · similarity 0.86)
- Articulate3D: Holistic Understanding of 3D Scenes as Universal Scene Description (ICCV 2025 · similarity 0.86)
- FreeMan: Towards Benchmarking 3D Human Pose Estimation under Real-World Conditions (CVPR 2024 · similarity 0.85)
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.
