Research map / 3D vision / 3D generation
SceneHGN: Hierarchical Graph Networks for 3D Indoor Scene Generation With Fine-Grained Geometry
Lin Gao, Jia-Mu Sun, Kaichun Mo, Yu‐Kun Lai, Leonidas J. Guibas, Jie 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 object-object edges (ablation) | SceneHGN · 3D-FRONT generation | FID: 145.398 -> 106.005 | Improved |
| add room-object edges (ablation) | SceneHGN · 3D-FRONT generation | FID: 150.478 -> 106.005 | Improved |
| add hyper-edges (ablation) | SceneHGN · 3D-FRONT generation | FID: 130.276 -> 106.005 | Improved |
| add hyper-edges (ablation) | SceneHGN · 3D-FRONT generation | o3: 1.553 -> 0.470 | Improved |
| add functional regions as intermediate level (ablation) | SceneHGN · 3D-FRONT reconstruction | CD (x10^-5): 398.1 -> 310.7 | Improved |
| add functional regions as intermediate level (ablation) | SceneHGN · 3D-FRONT generation | FID: 135.940 -> 106.005 | Improved |
| separate (multi-stage) training instead of end-to-end (ablation) | SceneHGN · 3D-FRONT reconstruction | CD (x10^-5): 473.9 -> 315.2 | Improved |
| separate (multi-stage) training instead of end-to-end (ablation) | SceneHGN · 3D-FRONT generation | FID: 170.079 -> 139.450 | Improved |
| jointly model layout and part geometry vs ATISS | SceneHGN vs ATISS · 3D-FRONT generation | FID: 141.7889 -> 139.4508 | Improved |
| hierarchical joint generation vs GRAINS | SceneHGN vs GRAINS · 3D-FRONT generation | o1: 0.30 -> 0.05 | Improved |
| SceneHGN vs GRAINS on orientation | SceneHGN vs GRAINS · 3D-FRONT generation | orientation score: 0.9931 -> 0.9572 | Hurt |
| SceneHGN vs ATISS on orientation | SceneHGN vs ATISS · 3D-FRONT generation | orientation score: 0.9578 -> 0.9572 | No clear effect |
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.
- ATISS — ATISS: Autoregressive Transformers for Indoor Scene Synthesis (NeurIPS 2021) · better on 4 of 5 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
- Hierarchically-Structured Open-Vocabulary Indoor Scene Synthesis with Pre-trained Large Language Model (AAAI 2025 · similarity 0.90)
- MMGDreamer: Mixed-Modality Graph for Geometry-Controllable 3D Indoor Scene Generation (AAAI 2025 · similarity 0.89)
- OptiScene: LLM-driven Indoor Scene Layout Generation via Scaled Human-aligned Data Synthesis and Multi-Stage Preference Optimization (NeurIPS 2025 · similarity 0.89)
- SceneCraft: Layout-Guided 3D Scene Generation (NeurIPS 2024 · similarity 0.89)
- DeBaRA: Denoising-Based 3D Room Arrangement Generation (NeurIPS 2024 · similarity 0.89)
- InstructScene: Instruction-Driven 3D Indoor Scene Synthesis with Semantic Graph Prior (ICLR 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.
