Research map / 3D vision / 3D generation
Dynamic-Robust Photometric–Semantic Reconstruction for Open-Vocabulary 3D Scene Understanding
Boyu Cai, Li Yang, Yan Xu, Wei Liu, Nian Liu, Sikui Zhang, Yan Wang, Chunfeng Yuan, Weiming Hu
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 dynamic-region-aware optimization (ablation) | baseline · D-RE10K (2 views, ablation) | PSNR: 15.66 -> 19.97 | Improved |
| add random masking (ablation) | unconstrained baseline · D-RE10K (2 views, ablation) | PSNR: 15.66 -> 17.10 | Improved |
| add targeted dynamic-region masking over random masking (ablation) | baseline + random masking · D-RE10K (2 views, ablation) | PSNR: 17.10 -> 19.97 | Improved |
| add dynamic-region-aware optimization (ablation) | baseline · D-RE10K (2 views, ablation) | LPIPS: 0.532 -> 0.339 | Improved |
| add semantic branch (ablation) | SPAR · D-RE10K (2 views, ablation) | PSNR: 19.77 -> 19.97 | Improved |
| add semantic branch (ablation) | SPAR · D-RE10K (2 views, ablation) | LPIPS: 0.356 -> 0.339 | Improved |
| add pre-training (ablation) | SPAR · D-RE10K (2 views, ablation) | PSNR: 18.66 -> 19.97 | Improved |
| add copy-paste augmentation (ablation) | SPAR · D-RE10K (2 views, ablation) | PSNR: 19.82 -> 19.97 | Improved |
| add test-time SAM2 refinement of the training-free pipeline (ablation) | SPAR · D-RE10K (2 views, ablation) | PSNR: 19.90 -> 19.97 | Improved |
| add SAM2 test-time mask refinement (ablation) | SPAR · D-RE10K motion mask (n=3 views) | mIoU: 75.7 -> 88.5 | Improved |
| add SAM2 test-time mask refinement (ablation) | SPAR · D-RE10K motion mask (n=2 views) | mIoU: 75.9 -> 87.8 | Improved |
| use SPAR over WildRayZer for NVS with 3 views | WildRayZer · D-RE10K (3 views) | PSNR: 21.98 -> 22.15 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- WildRayZer — WildRayZer: Self-supervised Large View Synthesis in Dynamic Environments (CVPR 2026) · better on 7 of 9 shared results
- SIU3R — SIU3R: Simultaneous Scene Understanding and 3D Reconstruction Beyond Feature Alignment (NeurIPS 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
- Know Your Neighbors: Improving Single-View Reconstruction via Spatial Vision-Language Reasoning (CVPR 2024 · similarity 0.89)
- Large Spatial Model: End-to-end Unposed Images to Semantic 3D (NeurIPS 2024 · similarity 0.89)
- Motion4D: Learning 3D-Consistent Motion and Semantics for 4D Scene Understanding (NeurIPS 2025 · similarity 0.89)
- Cross-Modal and Uncertainty-Aware Agglomeration for Open-Vocabulary 3D Scene Understanding (CVPR 2025 · similarity 0.88)
- Learning Temporal 3D Semantic Scene Completion via Optical Flow Guidance (NeurIPS 2025 · similarity 0.88)
- Tackling View-Dependent Semantics in 3D Language Gaussian Splatting (ICML 2025 · similarity 0.87)
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.
