AI Paper Atlas

Research map / 3D vision / 3D generation

SegVGGT: Joint 3D Reconstruction and Instance Segmentation from Multi-View Images

Jinyuan Qu, Hongyang Li, Lei Zhang

ECCV 2026 · 3D vision › 3D generation · approach: Learning 3D reconstruction from images · 3 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
add FADA attention-alignment loss (ablation)SegVGGT · ScanNet200 valmAP: 26.7 -> 31.1Improved
add FADA JS term to Hungarian matching cost (ablation)SegVGGT · ScanNet200 valmAP: 31.1 -> 31.9Improved
insert query interaction in all 24 layers instead of early 12 (ablation)SegVGGT · ScanNet200 valmAP: 23.8 -> 31.9Improved
insert query interaction in all 24 layers instead of late 12 (ablation)SegVGGT · ScanNet200 valmAP: 30.5 -> 31.9Improved
insert query interaction in all 24 layers instead of interleaved 12 (ablation)SegVGGT · ScanNet200 valmAP: 30.4 -> 31.9Improved
jointly fine-tune frame/global attention with LoRA instead of freezing (ablation)SegVGGT · ScanNet200 valmAP: 23.4 -> 31.9Improved
use unified architecture instead of separate 12-layer Mask2Former decoder on VGGT (ablation)SegVGGT · ScanNet200 valmAP: 23.4 -> 31.9Improved
increase separate mask decoder depth from 12 to 18 layers (ablation)VGGT(FT)+Mask Decoder · ScanNet200 valmAP: 23.4 -> 16.7Hurt
two-stage VGGT reconstruction then OneFormer3D instead of point cloud inputOneFormer3D · ScanNet200 valmAP: 30.2 -> 2.5Hurt
joint fine-tuning with instance segmentation vs geometry-only fine-tuningSegVGGT · ScanNet depth (10 scenes)Abs Rel: 0.0311 -> 0.0307No clear effect
disable PanSt3R QUBO post-processingPanSt3R · 100 input framestotal time (s): 381.96 -> 59.63Improved

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.

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