Research map / 3D vision / Depth and stereo
CasaMaestro: Multi-View Panoramas for House-Scale 3D Reconstruction
Yuzhou Ji, Xiaotian Yang, Zhipeng Zhang
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 panoramic cross-view pose decoder (vs linear head) (ablation) | CasaMaestro DINO-base · Realsee pose | AUC@30: 0.568 -> 0.831 | Improved |
| add panoramic cross-view pose decoder (vs linear head) (ablation) | CasaMaestro DINO-large · Realsee pose | AUC@30: 0.598 -> 0.871 | Improved |
| add ERP rotation augmentation (ablation) | CasaMaestro DINO-base, linear decoder · Realsee pose | AUC@30: 0.568 -> 0.772 | Improved |
| add ERP augmentation on top of pose decoder (ablation) | CasaMaestro DINO-large · Realsee pose | AUC@30: 0.871 -> 0.927 | Improved |
| add ERP augmentation (linear decoder) (ablation) | CasaMaestro DINO-base · Realsee depth | AbsRel: 0.092 -> 0.111 | Hurt |
| scale backbone from DINO-base to DINO-large (linear decoder, no aug) (ablation) | CasaMaestro · Realsee pose | AUC@30: 0.568 -> 0.598 | No clear effect |
| use panoramic pose decoder rather than ERP augmentation alone (ablation) | CasaMaestro · Realsee pose | AUC@30: decoder 0.831/0.871 vs aug 0.772/0.815 | Improved |
| panoramic multi-view model vs pinhole VFMs on sparse house panoramas | CasaMaestro · Realsee-Real | AUC@30: 0.510 -> 0.903 vs Pi3-Finetune | Improved |
| panoramic multi-view model vs best previous on synthetic | CasaMaestro · Realsee-Syn | AUC@30: 0.507 -> 0.927 vs Pi3-Finetune | Improved |
| fine-tune VGGT/Pi3 on Realsee training set | VGGT, Pi3 · Realsee pose | AUC@30: VGGT 0.388 -> 0.411; Pi3 0.450 -> 0.510 | Improved |
| multi-view panoramic depth vs single-view panoramic depth methods | CasaMaestro · zero-shot Replica/PanoSUNCG/HM3D | AbsRel: e.g. Replica 0.058 vs 0.070 best prior | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- PanoPose — PanoPose: Self-supervised Relative Pose Estimation for Panoramic Images (CVPR 2024) · better on 3 of 3 shared results
- PanoFormer — PanoFormer: Panorama Transformer for Indoor 360° Depth Estimation (ECCV 2022) · better on 3 of 3 shared results
- VGGT — VGGT: Visual Geometry Grounded Transformer (CVPR 2025) · better on 1 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
- MP-SfM: Monocular Surface Priors for Robust Structure-from-Motion (CVPR 2025 · similarity 0.85)
- Sparis: Neural Implicit Surface Reconstruction of Indoor Scenes from Sparse Views (AAAI 2025 · similarity 0.85)
- HouseLayout3D: A Benchmark and Training-free Baseline for 3D Layout Estimation in the Wild (NeurIPS 2025 · similarity 0.85)
- MvFFN: Multi-view Floor-Plan Feed-Forward Network for Unposed Wide-Baseline Panorama Layout Reconstruction (NeurIPS 2026 · similarity 0.85)
- PLANA3R: Zero-shot Metric Planar 3D Reconstruction via Feed-forward Planar Splatting (NeurIPS 2025 · similarity 0.85)
- SparseGNV: Generating Novel Views of Indoor Scenes with Sparse RGB-D Images (AAAI 2024 · similarity 0.84)
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.
