AI Paper Atlas

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

ECCV 2026 · 3D vision › 3D generation · approach: Learning 3D reconstruction from images · 0 citations (OpenAlex)

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 changedWhereMeasuredEffect
keep the view synthesis model frozen and propagate labels instead of joint 3D Gaussian trainingOurs vs SIU3R · ScanNetPSNR: 25.88 -> 33.56Improved
propagate labels through frozen NVS modelOurs vs SIU3R · ScanNet novel viewsPQ: 0.6565 -> 0.6092Hurt
low overlap evaluationOurs vs SIU3R · ScanNet low overlapPQ: 0.0473 -> 0.0011No clear effect
train NVS jointly with segmentation head (ablation)Ours · ScanNetPSNR: 33.56 -> 24.76Hurt
attach segmentation head to frozen decoder features (ablation)Ours · ScanNetPQ: 0.6092 -> 0.1816Hurt
add input-view segmentation plus propagation instead of segmenting rendered images (ablation)Ours · ScanNetPQ: 0.5740 -> 0.6092Improved
add uncertainty threshold tau=0.2 (ablation)Ours · ScanNet novel viewsPQ: 0.5449 -> 0.6092Improved
use binary encoding instead of grid encoding (ablation)Ours · ScanNet novel viewsPQ: 0.5270 -> 0.6092Improved
use binary encoding instead of HSV (ablation)Ours · ScanNet novel viewsPQ: 0.4436 -> 0.6092Improved
use Less3Depend instead of RayZer backbone (ablation)Ours · ScanNet novel viewsPQ: 0.5523 -> 0.6092Improved
use latent Plucker pose instead of GT pose with PanSt3ROurs (PanSt3R) · Replica zero-shotPQ: 0.246 -> 0.394Improved
increase input views from 2 to 3Ours · ScanNetPQ: 0.6092 -> 0.5711Hurt

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.

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

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy