Research map / 3D vision / 3D generation
Extending a Large View Synthesis Model for Multi-view Panoptic Segmentation
Kwonyoung Ryu, In-Jae Lee, Jonghyun Jin, Hyunjee Lee, Jongmin Lee, Jaesik Park
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 |
|---|---|---|---|
| keep the view synthesis model frozen and propagate labels instead of joint 3D Gaussian training | Ours vs SIU3R · ScanNet | PSNR: 25.88 -> 33.56 | Improved |
| propagate labels through frozen NVS model | Ours vs SIU3R · ScanNet novel views | PQ: 0.6565 -> 0.6092 | Hurt |
| low overlap evaluation | Ours vs SIU3R · ScanNet low overlap | PQ: 0.0473 -> 0.0011 | No clear effect |
| train NVS jointly with segmentation head (ablation) | Ours · ScanNet | PSNR: 33.56 -> 24.76 | Hurt |
| attach segmentation head to frozen decoder features (ablation) | Ours · ScanNet | PQ: 0.6092 -> 0.1816 | Hurt |
| add input-view segmentation plus propagation instead of segmenting rendered images (ablation) | Ours · ScanNet | PQ: 0.5740 -> 0.6092 | Improved |
| add uncertainty threshold tau=0.2 (ablation) | Ours · ScanNet novel views | PQ: 0.5449 -> 0.6092 | Improved |
| use binary encoding instead of grid encoding (ablation) | Ours · ScanNet novel views | PQ: 0.5270 -> 0.6092 | Improved |
| use binary encoding instead of HSV (ablation) | Ours · ScanNet novel views | PQ: 0.4436 -> 0.6092 | Improved |
| use Less3Depend instead of RayZer backbone (ablation) | Ours · ScanNet novel views | PQ: 0.5523 -> 0.6092 | Improved |
| use latent Plucker pose instead of GT pose with PanSt3R | Ours (PanSt3R) · Replica zero-shot | PQ: 0.246 -> 0.394 | Improved |
| increase input views from 2 to 3 | Ours · ScanNet | PQ: 0.6092 -> 0.5711 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SIU3R — SIU3R: Simultaneous Scene Understanding and 3D Reconstruction Beyond Feature Alignment (NeurIPS 2025) · better on 7 of 9 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
- Multi-view Consistent 3D Panoptic Scene Understanding (AAAI 2025 · similarity 0.87)
- Novel View Synthesis with Pixel-Space Diffusion Models (CVPR 2025 · similarity 0.87)
- LVSM: A Large View Synthesis Model with Minimal 3D Inductive Bias (ICLR 2025 · similarity 0.87)
- MultiDiff: Consistent Novel View Synthesis from a Single Image (CVPR 2024 · similarity 0.86)
- PanoRecon: Real-Time Panoptic 3D Reconstruction from Monocular Video (CVPR 2024 · similarity 0.86)
- GenRec: Knowing Where to Reconstruct and Where to Generate (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.
