Research map / 3D vision / 3D generation
PanelNet: Understanding 360 Indoor Environment via Panel Representation
Haozheng Yu, Lu He, Bing Jian, Weiwei Feng, Shan Liu
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 |
|---|---|---|---|
| Use PanelNet instead of the strongest listed baseline | indoor panorama depth estimation · Stanford2D3D | MRE: 0.1014 -> 0.0829 | Improved |
| Use PanelNet instead of the strongest listed baseline | indoor panorama depth estimation · Matterport3D | MRE: 0.1387 -> 0.1150 | Improved |
| Use PanelNet for RGB semantic segmentation | Stanford2D3D semantic segmentation · Stanford2D3D | mIoU: +3.0 | Improved |
| Use PanelNet for layout estimation | PanoContext layout estimation · PanoContext | 3DIoU (%): 85.16 -> 84.52 | Hurt |
| Add global panel geometry embedding (ablation) | ResNet-34 depth baseline · Stanford2D3D | MRE: 0.1033 -> 0.1029 | Improved |
| Add global and local panel geometry embedding (ablation) | ResNet-34 depth baseline · Stanford2D3D | MRE: 0.1033 -> 0.1000 | Improved |
| Add Panel Blocks (ablation) | ResNet-34 depth baseline · Stanford2D3D | MRE: 0.1033 -> 0.0904 | Improved |
| Add Window Blocks alongside Panel Blocks (ablation) | ResNet-34 depth baseline · Stanford2D3D | MRE: 0.0904 -> 0.0854 | Improved |
| Add local and global geometry embedding and Panel Blocks (ablation) | ResNet-34 depth baseline · Stanford2D3D | MRE: 0.1033 -> 0.0851 | Improved |
| Add Window Blocks to the geometry-embedding and Panel-Block model (ablation) | ResNet-34 depth baseline · Stanford2D3D | MRE: 0.0851 -> 0.0829 | Improved |
| Remove Local2Global Transformer (ablation) | PanelNet · Stanford2D3D | MRE: 0.0829 -> 0.1000 | Hurt |
| Replace tangent-image input with panel representation (ablation) | CNN-based panorama depth estimation · Stanford2D3D | MRE: 0.1132 -> 0.1000 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 on most shared results.
- Omnifusion — OmniFusion: 360 Monocular Depth Estimation via Geometry-Aware Fusion (CVPR 2022) · better on 3 of 3 shared results
- LGT-Net — LGT-Net: Indoor Panoramic Room Layout Estimation With Geometry-Aware Transformer Network (CVPR 2022) · better on 0 of 1 shared result
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
- LGT-Net: Indoor Panoramic Room Layout Estimation With Geometry-Aware Transformer Network (CVPR 2022 · similarity 0.89)
- AtlantaNet: Inferring the 3D Indoor Layout from a Single 360(∘) Image beyond the Manhattan World Assumption (ECCV 2020 · similarity 0.88)
- GeoLayout: Geometry Driven Room Layout Estimation Based on Depth Maps of Planes (ECCV 2020 · similarity 0.87)
- Geometric Exploitation for Indoor Panoramic Semantic Segmentation (NeurIPS 2024 · similarity 0.87)
- No More Ambiguity in 360deg Room Layout via Bi-Layout Estimation (CVPR 2024 · similarity 0.86)
- CAGE: Continuity-Aware edGE Network Unlocks Robust Floorplan Reconstruction (NeurIPS 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.
