Research map / 3D vision / Depth and stereo
PSMNet: Position-Aware Stereo Merging Network for Room Layout Estimation
Haiyan Wang, Will Hutchcroft, Yuguang Li, Zhiqiang Wan, Ivaylo Boyadzhiev, Yingli Tian, Sing Bing Kang
Check 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 PSMNet instead of LED²Net | ZInD stereo, w/ GT pose · ZInD stereo, Overall | 2D IoU (%): 76.39 -> 81.01 | Improved |
| Use PSMNet instead of LED²Net | ZInD stereo, w/o GT pose · ZInD stereo, Overall | 2D IoU (%): 65.81 -> 75.77 | Improved |
| Use PSMNet instead of LED²Net | ZInD stereo, w/o GT pose · Overlap-Low | 2D IoU (%): 63.14 -> 66.73 | Improved |
| Add CP² cross-perspective projection (ablation) | PSMNet, relative to the variant without CP², SP², and SE Attention · ZInD stereo, network-component ablation | 2D IoU (%): 60.41 -> 66.39 | Improved |
| Add SP² pose refinement (ablation) | PSMNet with CP² · ZInD stereo, network-component ablation | 2D IoU (%): 66.39 -> 70.92 | Improved |
| Add SE Attention (ablation) | PSMNet with CP² and SP² · ZInD stereo, network-component ablation | 2D IoU (%): 70.92 -> 72.38 | Improved |
| Add SP² pose refinement (ablation) | PSMNet with CP² and SE Attention · ZInD stereo, network-component ablation | 2D IoU (%): 69.14 -> 72.38 | Improved |
| Replace Mostly Manhattan post-processing with AtlantaNet post-processing (ablation) | PSMNet with GT pose · ZInD stereo | 2D IoU: 81.01 -> 77.69 | Hurt |
| Replace Mostly Manhattan post-processing with AtlantaNet post-processing (ablation) | PSMNet without GT pose · ZInD stereo | 2D IoU: 75.77 -> 73.22 | Hurt |
Later papers that compared with this method
12 later accepted papers report PSMNet as a baseline in their results tables; 11 of them report a better number on most of the results they share with it.
- Toward Ultrafast Depth Sensing via Active Event-Based Stereo Vision (TPAMI 2026 · better on 3 of 3 shared results)
- PolyLayout: Multi-room Manhattan Layout Estimation (ECCV 2026 · better on 1 of 1 shared result)
- HELVIPAD: A Real-World Dataset for Omnidirectional Stereo Depth Estimation (CVPR 2025 · better on 2 of 2 shared results)
- RobuSTereo: Robust Zero-Shot Stereo Matching under Adverse Weather (ICCV 2025 · better on 1 of 1 shared result)
- BANet: Bilateral Aggregation Network for Mobile Stereo Matching (ICCV 2025 · better on 1 of 1 shared result)
- Active Stereo in the Wild through Virtual Pattern Projection (IJCV 2025 · better on 1 of 1 shared result)
- Active Stereo Without Pattern Projector (ICCV 2023 · better on 1 of 1 shared result)
- FoggyStereo: Stereo Matching With Fog Volume Representation (CVPR 2022 · better on 5 of 5 shared results)
4 more not listed.
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
- LayoutNet: Reconstructing the 3D Room Layout From a Single RGB Image (CVPR 2018 · similarity 0.88)
- DuLa-Net: A Dual-Projection Network for Estimating Room Layouts From a Single RGB Panorama (CVPR 2019 · similarity 0.88)
- MvFFN: Multi-view Floor-Plan Feed-Forward Network for Unposed Wide-Baseline Panorama Layout Reconstruction (NeurIPS 2026 · similarity 0.87)
- LGT-Net: Indoor Panoramic Room Layout Estimation With Geometry-Aware Transformer Network (CVPR 2022 · similarity 0.87)
- SALVe: Semantic Alignment Verification for Floorplan Reconstruction from Sparse Panoramas (ECCV 2022 · similarity 0.86)
- PolyLayout: Multi-room Manhattan Layout Estimation (ECCV 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.
