AI Paper Atlas

Research map / 3D vision / 3D generation

ReGen3D: Generalizable Unified Representation Learning for 3D Understanding

Haotian Zhang, Jincen Jiang, Yuhang Li, Jian Jun Zhang, Meili Wang, Jian Chang

ECCV 2026 · 3D vision › 3D generation · approach: Semantic scene completion · 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
add URT (unified tokenization) to the baseline (ablation)ReGen3D baseline · Voxel held out, classificationaccuracy (%): 36.2 -> 51.8Improved
add URT (unified tokenization) to the baseline (ablation)ReGen3D baseline · Voxel held out, segmentationmIoU (%): 22.1 -> 52.4Improved
add URT to the baseline (ablation)ReGen3D baseline · Voxel held out, reconstructionCD (x10^-3): 57.2 -> 55.7No clear effect
add SMI (shared memory interaction) on top of URT (ablation)ReGen3D · Voxel held out, classificationaccuracy (%): 51.8 -> 72.6Improved
add SMI on top of URT (ablation)ReGen3D · Voxel held out, segmentationmIoU (%): 52.4 -> 58.6Improved
add SMI on top of URT (ablation)ReGen3D · Training representations P/M/G reconstructionCD (x10^-3): 27.2 -> 23.4 (Point)Improved
add ERA (efficient representation adapter) on top of URT and SMI (ablation)ReGen3D · Voxel held out, reconstructionCD (x10^-3): 50.6 -> 33.1Improved
add ERA on top of URT and SMI (ablation)ReGen3D · Voxel held out, segmentationmIoU (%): 58.6 -> 76.6Improved
add ERA on top of URT and SMI (ablation)ReGen3D · Training representations (P, M, G)reconstruction/classification/segmentation: 23.4 23.6 29.2 unchangedNo clear effect
replace representation-specific models with ReGen3D unified modelReGen3D vs ShapeSplat · RepShift-3D unseen Gaussian, classificationaccuracy (%): 45.2 -> 85.1Improved
replace naive cross-attention baseline with ReGen3DReGen3D vs Baseline · RepShift-3D unseen Mesh reconstructionCD (x10^-3): 52.4 -> 32.1Improved
use ERA to adapt to real-scan point cloudsReGen3D · ScanObjectNNaccuracy (%): 70.6 Acc., 40.1 CDNo clear effect

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