Research map / 3D vision / NeRF and radiance fields
RICO: Regularizing the Unobservable for Indoor Compositional Reconstruction
Zizhang Li, Xiaoyang Lyu, Yuanyuan Ding, Mengmeng Wang, Yiyi Liao, Yong 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 |
|---|---|---|---|
| add patch-based background smoothness loss L_bs (ablation) | RICO · Synthetic scenes (occluded background) | Depth error: 0.021 -> 0.004 | Improved |
| add object point-SDF loss L_op on top of L_bs (ablation) | RICO · Synthetic scenes (objects) | Chamfer-L1: 1.25 -> 0.045 | Improved |
| add reversed depth loss L_rd on top of L_bs and L_op (ablation) | RICO · Synthetic scenes (objects) | F-score: 0.735 -> 0.817 | Improved |
| add background smoothness loss when object losses are already used (ablation) | RICO · Synthetic scenes (objects) | F-score: 0.785 -> 0.817 | Improved |
| add L_bs alone without object regularizers (ablation) | RICO · Synthetic scenes (objects) | F-score: 0.065 -> 0.066 | No clear effect |
| set epsilon threshold to 0.05 instead of 0 in object point-SDF loss (ablation) | RICO · Synthetic scenes (objects) | Chamfer-L1: 0.187 -> 0.033 | Improved |
| increase epsilon from 0.05 to 0.2 (ablation) | RICO · Synthetic scenes (objects) | F-score: 0.817 -> 0.793 | Hurt |
| use NeuS rendering scheme instead of VolSDF (ablation) | RICO · Synthetic scenes (objects) | F-score: 0.751 -> 0.817 | Improved |
| RICO regularization versus ObjSDF* baseline on real ScanNet visible-surface metrics | RICO · ScanNet (7 scenes, whole scene) | Chamfer-L1: 0.092 -> 0.088 | Improved |
| RICO versus ObjSDF* on 2D rendering and segmentation | RICO · ScanNet | PSNR: 23.15 -> 23.30 | No clear effect |
| add post-process culling of out-of-room parts (ObjSDF*-C) versus RICO regularization | ObjSDF* · Synthetic scenes (objects) | F-score: 0.085 -> 0.681 | Improved |
Later papers that compared with this method
1 later accepted paper reports RICO as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- PhyRecon: Physically Plausible Neural Scene Reconstruction (NeurIPS 2024 · better on 3 of 3 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
- ObjectSDF++: Improved Object-Compositional Neural Implicit Surfaces (ICCV 2023 · similarity 0.87)
- Pseudo-Plane Regularized Signed Distance Field for Neural Indoor Scene Reconstruction (IJCV 2024 · similarity 0.87)
- DebSDF: Delving Into the Details and Bias of Neural Indoor Scene Reconstruction (TPAMI 2024 · similarity 0.86)
- NC-SDF: Enhancing Indoor Scene Reconstruction Using Neural SDFs with View-Dependent Normal Compensation (CVPR 2024 · similarity 0.85)
- Behind the Veil: Enhanced Indoor 3D Scene Reconstruction with Occluded Surfaces Completion (CVPR 2024 · similarity 0.85)
- MOVIS: Enhancing Multi-Object Novel View Synthesis for Indoor Scenes (CVPR 2025 · 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.
