MERG3R: A Divide-and-Conquer Approach to Large-Scale Neural Visual Geometry

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Cheng, Leo Kaixuan, Shaikh, Abdus, Liang, Ruofan, Wu, Zhijie, Guan, Yushi, Vijaykumar, Nandita
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914364539273216
author Cheng, Leo Kaixuan
Shaikh, Abdus
Liang, Ruofan
Wu, Zhijie
Guan, Yushi
Vijaykumar, Nandita
author_facet Cheng, Leo Kaixuan
Shaikh, Abdus
Liang, Ruofan
Wu, Zhijie
Guan, Yushi
Vijaykumar, Nandita
contents Recent advancements in neural visual geometry, including transformer-based models such as VGGT and Pi3, have achieved impressive accuracy on 3D reconstruction tasks. However, their reliance on full attention makes them fundamentally limited by GPU memory capacity, preventing them from scaling to large, unordered image collections. We introduce MERG3R, a training-free divide-and-conquer framework that enables geometric foundation models to operate far beyond their native memory limits. MERG3R first reorders and partitions unordered images into overlapping, geometrically diverse subsets that can be reconstructed independently. It then merges the resulting local reconstructions through an efficient global alignment and confidence-weighted bundle adjustment procedure, producing a globally consistent 3D model. Our framework is model-agnostic and can be paired with existing neural geometry models. Across large-scale datasets, including 7-Scenes, NRGBD, Tanks & Temples, and Cambridge Landmarks, MERG3R consistently improves reconstruction accuracy, memory efficiency, and scalability, enabling high-quality reconstruction when the dataset exceeds memory capacity limits.
format Preprint
id arxiv_https___arxiv_org_abs_2603_02351
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle MERG3R: A Divide-and-Conquer Approach to Large-Scale Neural Visual Geometry
Cheng, Leo Kaixuan
Shaikh, Abdus
Liang, Ruofan
Wu, Zhijie
Guan, Yushi
Vijaykumar, Nandita
Computer Vision and Pattern Recognition
Recent advancements in neural visual geometry, including transformer-based models such as VGGT and Pi3, have achieved impressive accuracy on 3D reconstruction tasks. However, their reliance on full attention makes them fundamentally limited by GPU memory capacity, preventing them from scaling to large, unordered image collections. We introduce MERG3R, a training-free divide-and-conquer framework that enables geometric foundation models to operate far beyond their native memory limits. MERG3R first reorders and partitions unordered images into overlapping, geometrically diverse subsets that can be reconstructed independently. It then merges the resulting local reconstructions through an efficient global alignment and confidence-weighted bundle adjustment procedure, producing a globally consistent 3D model. Our framework is model-agnostic and can be paired with existing neural geometry models. Across large-scale datasets, including 7-Scenes, NRGBD, Tanks & Temples, and Cambridge Landmarks, MERG3R consistently improves reconstruction accuracy, memory efficiency, and scalability, enabling high-quality reconstruction when the dataset exceeds memory capacity limits.
title MERG3R: A Divide-and-Conquer Approach to Large-Scale Neural Visual Geometry
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.02351