Research map / 3D vision / 3D generation
PUF: Plug-and-Play Uncertainty-Aware Fusion for Online 3D Scene Graph Generation
Yi Yang, Myrna Castillo Silva, Bodo Rosenhahn, Michael Yang
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 Dirichlet node representation with probabilistic association (ablation) | PUF-Gaussian · 3DSSG test | relationship recall (%): 27.9 -> 31.3 | Improved |
| add Dirichlet edge representation on top of node Dirichlet (ablation) | PUF-Gaussian · 3DSSG test | relationship recall (%): 31.3 -> 33.9 | Improved |
| add Dirichlet edge representation alone, without node Dirichlet (ablation) | PUF-Gaussian · 3DSSG test | relationship recall (%): 27.9 -> 27.9 | No clear effect |
| add relationship prior on top of node and edge Dirichlet (ablation) | PUF-Gaussian · 3DSSG test | relationship recall (%): 33.9 -> 46.0 | Improved |
| add Dirichlet node representation to edge plus prior (ablation) | PUF-Gaussian · 3DSSG test | relationship recall (%): 41.4 -> 46.0 | Improved |
| add Dirichlet node representation (ablation) | PUF-Voxel · 3DSSG test | relationship recall (%): 23.7 -> 28.6 | Improved |
| add relationship prior on top of node and edge Dirichlet (ablation) | PUF-Voxel · 3DSSG test | relationship recall (%): 31.3 -> 40.3 | Improved |
| add edge Dirichlet on top of node Dirichlet, no prior (ablation) | PUF-Gaussian · ReplicaSSG test | relationship recall (%): 23.1 -> 25.3 | Improved |
| replace frequentist prior with probabilistic Bayesian prior (ablation) | PUF-Gaussian · 3DSSG test | relationship recall (%): 43.1 -> 46.0 | Improved |
| add frequentist prior to FROSS (ablation) | FROSS-Gaussian · 3DSSG test | relationship recall (%): 27.9 -> 36.7 | Improved |
| raise semantic bandwidth sigma_se from 0.3 to 0.4 (ablation) | PUF-Gaussian · 3DSSG validation | relationship recall (%): 34.2 -> 29.3 | Hurt |
| lower semantic bandwidth sigma_se from 0.3 to 0.2 (ablation) | PUF-Gaussian · 3DSSG validation | relationship recall (%): 34.2 -> 32.4 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- FROSS — FROSS: Faster-Than-Real-Time Online 3D Semantic Scene Graph Generation from RGB-D Images (ICCV 2025) · better on 7 of 8 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- DepWorldSG: Depth-Aware 3D Semantic Scene Graph Generation via World-Model Priors (ECCV 2026 · similarity 0.91)
- GraphWrit3R: End-to-End Writing of Scene Graphs for Multi-Modal 3D Scenes (NeurIPS 2026 · similarity 0.89)
- FunFact: Building Probabilistic Functional 3D Scene Graphs via Factor-Graph Reasoning (CVPR 2026 · similarity 0.89)
- Statistical Confidence Rescoring for Robust 3D Scene Graph Generation from Multi-View Images (ICCV 2025 · similarity 0.89)
- Conformal Prediction and MLLM aided Uncertainty Quantification in Scene Graph Generation (CVPR 2025 · similarity 0.87)
- Cross-Modal and Uncertainty-Aware Agglomeration for Open-Vocabulary 3D Scene Understanding (CVPR 2025 · similarity 0.87)
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.
