Research map / 3D vision / 3D generation
DeepPanoContext: Panoramic 3D Scene Understanding With Holistic Scene Context Graph and Relation-Based Optimization
Cheng Zhang, Zhaopeng Cui, Cai Chen, Shuaicheng Liu, Bing Zeng, Hujun Bao, Yinda 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 relation optimization (ablation) | Ours · iGibson panorama (57 categories) | 3D detection mAP: 30.91 -> 33.59 | Improved |
| add relation optimization (ablation) | Ours · iGibson panorama | avg collisions per scene: 2.68 -> 0.86 | Improved |
| add relation features (Fr) (ablation) | Ours · iGibson panorama (57 categories) | 3D detection mAP: 33.22 -> 33.59 | Improved |
| add object features (Fo) (ablation) | Ours · iGibson panorama (57 categories) | 3D detection mAP: 33.32 -> 33.59 | Improved |
| add collision term Ec in RO (ablation) | Ours · iGibson panorama | avg collisions per scene: 2.05 -> 0.86 | Improved |
| add relation term Er in RO (ablation) | Ours · iGibson panorama (57 categories) | 3D detection mAP: 29.88 -> 33.59 | Improved |
| add observation term Eo in RO (ablation) | Ours · iGibson panorama (57 categories) | 3D detection mAP: 25.30 -> 33.59; collisions 0.09 -> 0.86 | Mixed |
| apply relation optimization to Total3D | Total3D · iGibson panorama (57 categories) | 3D detection mAP: 25.79 -> 32.46 | Improved |
| apply relation optimization to Im3D | Im3D · iGibson panorama (57 categories) | 3D detection mAP: 27.25 -> 33.54 | Improved |
| use full 360 degree FoV instead of 30 degree splits (full model) (ablation) | Ours · iGibson panorama (57 categories) | 3D detection mAP: 18.52 -> 33.92 | Improved |
| use full 360 degree FoV instead of 30 degree splits (w/o RO) (ablation) | Ours w/o RO · iGibson panorama (57 categories) | 3D detection mAP: 25.51 -> 30.91 | Improved |
| add RGCN relation-based context versus Im3D-Pano | Ours w/o RO · iGibson panorama (11 categories) | 3D detection mAP: 44.34 -> 50.68 | Improved |
Closest papers by meaning
- PanoContext-Former: Panoramic Total Scene Understanding with a Transformer (CVPR 2024 · similarity 0.93)
- Scene4U: Hierarchical Layered 3D Scene Reconstruction from Single Panoramic Image for Your Immerse Exploration (CVPR 2025 · similarity 0.87)
- HUSH: Holistic Panoramic 3D Scene Understanding using Spherical Harmonics (CVPR 2025 · similarity 0.87)
- Geometric Exploitation for Indoor Panoramic Semantic Segmentation (NeurIPS 2024 · similarity 0.87)
- Unified Panoramic Geometry Estimation via Multi-View Foundation Models (NeurIPS 2026 · similarity 0.86)
- PanoWorld: Towards Spatial Supersensing in 360◦ Panorama World (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.
