Research map / 3D vision / 3D generation
NoPA: Non-Parametric Online 3D Scene Graph Generation
Qi Xun Yeo, Seungjun Lee, Yan Li, Gim Hee Lee
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 |
|---|---|---|---|
| replace Gaussian with non-parametric particle set (FROSS merging kept) (ablation) | FROSS · 3DSSG (online, test) | relationship recall (%): 25.7 -> 17.6 | Hurt |
| replace Gaussian with non-parametric particle set (FROSS merging kept) (ablation) | FROSS · 3DSSG (online, test) | object recall (%): 60.6 -> 66.1 | Improved |
| add MMD-based merging on top of particle set (ablation) | NoPA · 3DSSG (online, test) | object recall (%): 66.1 -> 69.0 | Improved |
| add MMD-based merging on top of particle set (ablation) | NoPA · 3DSSG (online, test) | relationship recall (%): 17.6 -> 26.3 | Improved |
| add relationship propagation (ablation) | NoPA · 3DSSG (online, test) | relationship recall (%): 26.3 -> 53.2 | Improved |
| add relationship propagation (ablation) | NoPA · 3DSSG (online, test) | predicate recall (%): 31.0 -> 61.4 | Improved |
| raise MMD threshold from 0.6 to 0.7 (ablation) | NoPA · 3DSSG (online, test) val | relationship recall (%): 36.3 -> 53.7 | Improved |
| raise MMD threshold from 0.7 to 0.9 (ablation) | NoPA · 3DSSG (online, test) val | object recall (%): 66.0 -> 59.2 | Hurt |
| raise MMD threshold from 0.7 to 0.9 (ablation) | NoPA · 3DSSG (online, test) val | relationship recall (%): 53.7 -> 46.3 | Hurt |
| use margin band with MMD (epsilon 0.05) instead of Hellinger only (epsilon 0) (ablation) | NoPA · ReplicaSSG | relationship recall (%): 16.2 -> 36.9 | Improved |
| use relationship propagation instead of merging to aggregate relations (ablation) | NoPA · 3DSSG (online, test) | relationship recall (%): 25.2 -> 53.2 | Improved |
| use particle set instead of 3D bounding box (ablation) | NoPA · 3DSSG (online, test) | relationship recall (%): 7.0 -> 53.2 | Improved |
Closest papers by meaning
- PUF: Plug-and-Play Uncertainty-Aware Fusion for Online 3D Scene Graph Generation (ECCV 2026 · similarity 0.89)
- GraphWrit3R: End-to-End Writing of Scene Graphs for Multi-Modal 3D Scenes (NeurIPS 2026 · similarity 0.88)
- Points-to-3D: Structure-Aware 3D Generation with Point Cloud Priors (CVPR 2026 · similarity 0.88)
- FROSS: Faster-Than-Real-Time Online 3D Semantic Scene Graph Generation from RGB-D Images (ICCV 2025 · similarity 0.88)
- Open3DSG: Open-Vocabulary 3D Scene Graphs from Point Clouds with Queryable Objects and Open-Set Relationships (CVPR 2024 · similarity 0.87)
- Generative Gaussian Splatting for Unbounded 3D City Generation (CVPR 2025 · similarity 0.86)
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.
