Research map / 3D vision / 3D generation
OpenRooms: An Open Framework for Photorealistic Indoor Scene Datasets
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Train the inverse-rendering network on OpenRooms rather than the listed prior synthetic datasets | Intrinsic decomposition network · IIW | WHDR: 16.4; Li20: 15.9 | Mixed |
| Train the inverse-rendering network on OpenRooms | Normal and depth estimation network · NYU | Depth error: 0.184 -> 0.171 | Helped |
| Train the light-source detector on OpenRooms in addition to NYU2 | Light source detector · NYU2 light source detection | AP(0.5:0.95), bbox: 17.1 -> 23.5 | Helped |
| Pre-train the segmentation model on OpenRooms rather than InteriorNet | DeepLabV3 · NYUv2 semantic segmentation, 50K images | mIoU: 42.2 -> 42.9 | Helped |
| Add cascade refinement (ablation) | Inverse-rendering network · OpenRooms inverse rendering test set | Albedo scale-invariant L2 loss (10^-3): 9.99 -> 9.43 | Helped |
| Add cascade refinement (ablation) | Inverse-rendering network · OpenRooms inverse rendering test set | Normal L2 loss (10^-2): 4.51 -> 4.42 | Helped |
| Add cascade refinement (ablation) | Inverse-rendering network · OpenRooms inverse rendering test set | Depth scale-invariant log L2 loss (10^-2): 5.18 -> 4.89 | Helped |
| Add cascade refinement (ablation) | Inverse-rendering network · OpenRooms inverse rendering test set | Roughness L2 loss (10^-2): 6.59 -> 6.64 | Hurt |
| Add cascade refinement (ablation) | Inverse-rendering network · OpenRooms inverse rendering test set | Per-pixel lighting scale-invariant log(x+1) L2 loss: 0.150 -> 0.146 | Helped |
| Add a bilateral solver (ablation) | Inverse-rendering network · OpenRooms inverse rendering test set | Albedo scale-invariant L2 loss (10^-3): 9.43 -> 9.29 | Helped |
| Add a bilateral solver (ablation) | Inverse-rendering network · OpenRooms inverse rendering test set | Depth scale-invariant log L2 loss (10^-2): 4.89 -> 4.86 | Helped |
| Use OpenRooms-trained lighting predictions for object insertion | Object insertion system · Real-image object insertion user study | Pairwise preference over prior methods: 88.19%, 66.16%, 56.53%, 54.77% | Helped |
Closest papers by meaning
- Inverse Rendering for Complex Indoor Scenes: Shape, Spatially-Varying Lighting and SVBRDF From a Single Image (CVPR 2020 · similarity 0.90)
- LiteReality: Graphic-Ready 3D Scene Reconstruction from RGB-D Scans (NeurIPS 2025 · similarity 0.86)
- InternScenes: A Large-scale Simulatable Indoor Scene Dataset with Realistic Layouts (NeurIPS 2025 · similarity 0.86)
- Intrinsic Image Diffusion for Indoor Single-view Material Estimation (CVPR 2024 · similarity 0.85)
- SparseGNV: Generating Novel Views of Indoor Scenes with Sparse RGB-D Images (AAAI 2024 · similarity 0.85)
- S-INF: Towards Realistic Indoor Scene Synthesis via Scene Implicit Neural Field (AAAI 2025 · 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.