Research map / 3D vision / Depth and stereo
Joint 3D Layout and Depth Prediction from a Single Indoor Panorama Image
Wei Zeng, Sezer Karaoglu, Theo Gevers
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 predicted depth as input to layout depth network (ablation) | layout depth network · Stanford 2D-3D layout | 3D IoU (%): 77.28 -> 82.65 | Improved |
| add predicted semantics as input to layout depth network (ablation) | layout depth network · Stanford 2D-3D layout | 3D IoU (%): 77.28 -> 78.57 | Improved |
| add both depth and semantics as input (ablation) | layout depth network · Stanford 2D-3D layout | 3D IoU (%): 77.28 -> 85.81 | Improved |
| add virtual normal loss (ablation) | layout depth network · Stanford 2D-3D layout | 3D IoU (%): 84.22 -> 85.81 | Improved |
| add virtual normal loss (ablation) | layout depth network · Stanford 2D-3D layout | corner error (%): 0.75 -> 0.67 | Improved |
| predict layout depth map instead of edge and corner maps (ablation) | layout network · Stanford 2D-3D layout | 3D IoU (%): 82.03 -> 85.81 | Improved |
| predict layout depth map instead of edge/corner maps converted from layout depth (ablation) | layout network · Stanford 2D-3D layout | 3D IoU (%): 83.67 -> 85.81 | Improved |
| add layout-depth refinement (direct refinement) (ablation) | depth network · Stanford 2D-3D depth | Abs Rel: 0.105 -> 0.089 | Improved |
| use semantic-guided fusion instead of direct refinement (ablation) | depth network · Stanford 2D-3D depth | Abs Rel: 0.089 -> 0.086 | Improved |
| add final refinement after semantic-guided fusion (ablation) | depth network · Stanford 2D-3D depth | Abs Rel: 0.086 -> 0.068 | Improved |
| train coarse depth module jointly (ablation) | coarse depth · Stanford 2D-3D depth | Abs Rel: 0.112 -> 0.105 | Improved |
| train depth refinement jointly (ablation) | depth refinement · Stanford 2D-3D depth | Abs Rel: 0.084 -> 0.068 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- HorizonNet — HorizonNet: Learning Room Layout With 1D Representation and Pano Stretch Data Augmentation (CVPR 2019) · better on 5 of 5 shared results
- LayoutNet — LayoutNet: Reconstructing the 3D Room Layout From a Single RGB Image (CVPR 2018) · better on 5 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
- LED2-Net: Monocular 360deg Layout Estimation via Differentiable Depth Rendering (CVPR 2021 · similarity 0.90)
- SSLayout360: Semi-Supervised Indoor Layout Estimation From 360deg Panorama (CVPR 2021 · similarity 0.89)
- Layout-Guided Novel View Synthesis From a Single Indoor Panorama (CVPR 2021 · similarity 0.89)
- 3D Room Layout Estimation from a Cubemap of Panorama Image via Deep Manhattan Hough Transform (ECCV 2022 · similarity 0.89)
- HouseLayout3D: A Benchmark and Training-free Baseline for 3D Layout Estimation in the Wild (NeurIPS 2025 · similarity 0.87)
- MvFFN: Multi-view Floor-Plan Feed-Forward Network for Unposed Wide-Baseline Panorama Layout Reconstruction (NeurIPS 2026 · 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.
