Research map / 3D vision / NeRF and radiance fields
Object-Compositional Neural Implicit Surfaces
Qianyi Wu, Xian Liu, Yuedong Chen, Kejie Li, Chuanxia Zheng, Jianfei Cai, Jianmin Zheng
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 |
|---|---|---|---|
| model per-object SDFs with semantics as function of object SDF (vs scene SDF plus semantic head) (ablation) | VolSDF · ToyDesk | Chamfer distance: 0.34 -> 0.19 | Improved |
| model per-object SDFs with semantics as function of object SDF (ablation) | VolSDF w/ Semantic · ScanNet | Chamfer distance: 0.28 -> 0.22 | Improved |
| model per-object SDFs with semantics as function of object SDF (ablation) | VolSDF · ToyDesk | PSNR (dB): 22.00 -> 22.00 | No clear effect |
| model per-object SDFs vs semantic head (ablation) | VolSDF w/ Semantic · ToyDesk | mIOU: 0.89 -> 0.88 | No clear effect |
| model per-object SDFs vs semantic head (ablation) | VolSDF w/ Semantic · ScanNet | mIOU: 0.56 -> 0.53 | Hurt |
| add semantic head to VolSDF (ablation) | VolSDF · ToyDesk | Chamfer distance: 0.30 -> 0.34 | Hurt |
| add semantic head to VolSDF (ablation) | VolSDF · ScanNet | Chamfer distance: 0.32 -> 0.28 | Improved |
| ObjectSDF instead of ObjectNeRF | ObjectNeRF · ToyDesk | Chamfer distance: 1.06 -> 0.19 | Improved |
| ObjectSDF instead of ObjectNeRF | ObjectNeRF · ScanNet | mIOU: 0.31 -> 0.53 | Improved |
| ObjectSDF instead of SemanticNeRF | SemanticNeRF · ScanNet | mIOU: 0.57 -> 0.53 | Hurt |
| ObjectSDF instead of ObjectNeRF (parameter count) | ObjectNeRF · parameters | #params: 1.78M(+19.20M) -> 0.804M | Improved |
| use threshold on semantic output to extract instances (ablation) | VolSDF w/ Semantic · instance extraction | instance IoU | Hurt |
Closest papers by meaning
- ObjectSDF++: Improved Object-Compositional Neural Implicit Surfaces (ICCV 2023 · similarity 0.94)
- ClusteringSDF: Self-Organized Neural Implicit Surfaces for 3D Decomposition (ECCV 2024 · similarity 0.88)
- Towards Unbiased Volume Rendering of Neural Implicit Surfaces With Geometry Priors (CVPR 2023 · similarity 0.88)
- NeuralUDF: Learning Unsigned Distance Fields for Multi-View Reconstruction of Surfaces With Arbitrary Topologies (CVPR 2023 · similarity 0.88)
- Neural Unsigned Distance Fields for Implicit Function Learning (NeurIPS 2020 · similarity 0.87)
- S2MDF: A Plug-And-Play Layer for Intersection-Free Multi-Object Signed Distance Fields (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.
