SceneGraphGrounder: Zero-Shot 3D Visual Grounding via Structured Scene Graph Matching

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Sun, Xuefei, Zhang, Xujia, Crowe, Brendan, Albin, Doncey, Heckman, Christoffer
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916033426620416
author Sun, Xuefei
Zhang, Xujia
Crowe, Brendan
Albin, Doncey
Heckman, Christoffer
author_facet Sun, Xuefei
Zhang, Xujia
Crowe, Brendan
Albin, Doncey
Heckman, Christoffer
contents Zero-shot 3D visual grounding requires localizing objects in unstructured environments from free-form natural language. Recent vision-language model (VLM) approaches achieve promising results but rely on view-dependent reasoning or implicit representations, limiting spatial consistency and interpretability for compositional queries. We propose SceneGraphGrounder, a framework that reformulates 3D grounding as structured graph matching over a reconstructed 3D scene graph. To enable this formulation, we introduce a visual marker prompting strategy that enables a VLM to infer object-object relationships from 2D views, which are subsequently lifted into a persistent 3D scene graph encoding both spatial and semantic relations. Given a query, we construct a query graph and perform constrained alignment with the scene graph, ensuring multi-view consistency and interpretable reasoning. Experiments on the ScanRefer benchmark demonstrate that our method achieves competitive performance among zero-shot approaches, using only RGB-D inputs. We further validate our framework through real-world deployment on a mobile robot, demonstrating robust spatial reasoning in long-horizon physical environments. We will make our code publicly available upon acceptance.
format Preprint
id arxiv_https___arxiv_org_abs_2605_21788
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle SceneGraphGrounder: Zero-Shot 3D Visual Grounding via Structured Scene Graph Matching
Sun, Xuefei
Zhang, Xujia
Crowe, Brendan
Albin, Doncey
Heckman, Christoffer
Computer Vision and Pattern Recognition
Robotics
Zero-shot 3D visual grounding requires localizing objects in unstructured environments from free-form natural language. Recent vision-language model (VLM) approaches achieve promising results but rely on view-dependent reasoning or implicit representations, limiting spatial consistency and interpretability for compositional queries. We propose SceneGraphGrounder, a framework that reformulates 3D grounding as structured graph matching over a reconstructed 3D scene graph. To enable this formulation, we introduce a visual marker prompting strategy that enables a VLM to infer object-object relationships from 2D views, which are subsequently lifted into a persistent 3D scene graph encoding both spatial and semantic relations. Given a query, we construct a query graph and perform constrained alignment with the scene graph, ensuring multi-view consistency and interpretable reasoning. Experiments on the ScanRefer benchmark demonstrate that our method achieves competitive performance among zero-shot approaches, using only RGB-D inputs. We further validate our framework through real-world deployment on a mobile robot, demonstrating robust spatial reasoning in long-horizon physical environments. We will make our code publicly available upon acceptance.
title SceneGraphGrounder: Zero-Shot 3D Visual Grounding via Structured Scene Graph Matching
topic Computer Vision and Pattern Recognition
Robotics
url https://arxiv.org/abs/2605.21788