# Graph3R: Unifying View Graph Initialization with Feed-Forward Reconstruction

arXiv: https://arxiv.org/abs/2610.16272 · HTML: https://mlsanity.com/paper/2610.16272/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 26 of its 30 references that are in the map.
- First posted on arXiv 12 Oct 2026.

## Builds on (most cited first, 26 of 26)

- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Supervised Contrastive Learning](https://mlsanity.com/paper/2004.11362/) (2020)
- [VGGT: Visual Geometry Grounded Transformer](https://mlsanity.com/paper/2503.11651/) (2025)
- [ORB-SLAM: a Versatile and Accurate Monocular SLAM System](https://mlsanity.com/paper/1502.00956/) (2015)
- [Grounding Image Matching in 3D with MASt3R](https://mlsanity.com/paper/2406.09756/) (2024)
- [ORB-SLAM3: An Accurate Open-Source Library for Visual, Visual-Inertial and Multi-Map SLAM](https://mlsanity.com/paper/2007.11898/) (2020)
- [MVSNet: Depth Inference for Unstructured Multi-view Stereo](https://mlsanity.com/paper/1804.02505/) (2018)
- [π3: Permutation-Equivariant Visual Geometry Learning](https://mlsanity.com/paper/2507.13347/) (2025)
- [Global Structure-from-Motion Revisited](https://mlsanity.com/paper/2407.20219/) (2024)
- [MASt3R-SLAM: Real-Time Dense SLAM with 3D Reconstruction Priors](https://mlsanity.com/paper/2412.12392/) (2024)
- [Rethinking Visual Geo-localization for Large-Scale Applications](https://mlsanity.com/paper/2204.02287/) (2022)
- [AnyLoc: Towards Universal Visual Place Recognition](https://mlsanity.com/paper/2308.00688/) (2023)
- [VGGT-SLAM: Dense RGB SLAM Optimized on the SL(4) Manifold](https://mlsanity.com/paper/2505.12549/) (2025)
- [GSV-Cities: Toward Appropriate Supervised Visual Place Recognition](https://mlsanity.com/paper/2210.10239/) (2022)
- [MegaLoc: One Retrieval to Place Them All](https://mlsanity.com/paper/2502.17237/) (2025)
- [RoMa v2: Harder Better Faster Denser Feature Matching](https://mlsanity.com/paper/2511.15706/) (2025)
- [Doppelgangers: Learning to Disambiguate Images of Similar Structures](https://mlsanity.com/paper/2309.02420/) (2023)
- [360Loc: A Dataset and Benchmark for Omnidirectional Visual Localization with Cross-device Queries](https://mlsanity.com/paper/2311.17389/) (2023)
- [Emergent Outlier View Rejection in Visual Geometry Grounded Transformers](https://mlsanity.com/paper/2512.04012/) (2025)
- [Doppelgangers++: Improved Visual Disambiguation with Geometric 3D Features](https://mlsanity.com/paper/2412.05826/) (2024)
- [GeoVideo: Introducing Geometric Regularization into Video Generation Model](https://mlsanity.com/paper/2512.03453/) (2025)
- [GV-Bench: Benchmarking Local Feature Matching for Geometric Verification of Long-term Loop Closure Detection](https://mlsanity.com/paper/2407.11736/) (2024)
- [Global Structure-from-Motion Meets Feedforward Reconstruction](https://mlsanity.com/paper/2605.26103/) (2026)
- [Condition-Invariant and Compact Visual Place Description by Convolutional Autoencoder](https://mlsanity.com/paper/2204.07350/) (2022)
- [MR.ScaleMaster: Scale-Consistent Collaborative Mapping from Crowd-Sourced Monocular Videos](https://mlsanity.com/paper/2604.11372/) (2026)
- [3D Foundation Model-Based Loop Closing for Decentralized Collaborative SLAM](https://mlsanity.com/paper/2602.02430/) (2026)

## Abstract (arXiv)

Feed-forward reconstruction recovers camera poses and dense geometry from image sets whose views are assumed to come from a single scene. In practice, uncurated image collections often contain views from disjoint scenes, and reconstructing these views jointly can corrupt camera poses and geometry. Existing pipelines typically resolve view associations in separate stages before final reconstruction, leaving view graph initialization and reconstruction inference only loosely coupled. We introduce Graph3R, which unifies view graph initialization with feed-forward reconstruction through a shared geometric representation. A lightweight verification head predicts pairwise covisibility from the multi-view features of π3 to construct a view graph. This graph classifies the input views into scene groups. For each group, we mask the shared features before decoding camera poses and dense geometry. Reusing the multi-view features allows Graph3R to reconstruct disjoint scenes separately with a single backbone pass. Across three benchmarks, Graph3R improves average maximum recall at 100% precision by upto 8.99 percentage points over the baseline, while improving reconstruction quality on mixed-scene inputs. Regarding the multi-session mapping task, Graph3R improves cross-session localization and achieves competitive pose estimation accuracy, while delivering a runtime speedup of approximately 35.9× over the baseline when reconstructing the query group. By integrating learned view association with geometric inference, our approach takes a step toward end-to-end reconstruction from uncurated image collections.