mlsanityData to 7 Oct 2026

SG-PGM: Partial Graph Matching Network with Semantic Geometric Fusion for 3D Scene Graph Alignment and Its Downstream Tasks

arXiv 2403.19474 ↗First posted 28 Mar 2024cs.CV cs.ROcited by 10

Not on the map. Why?

Abstract (arXiv)

Scene graphs have been recently introduced into 3D spatial understanding as a comprehensive representation of the scene. The alignment between 3D scene graphs is the first step of many downstream tasks such as scene graph aided point cloud registration, mosaicking, overlap checking, and robot navigation. In this work, we treat 3D scene graph alignment as a partial graph-matching problem and propose to solve it with a graph neural network. We reuse the geometric features learned by a point cloud registration method and associate the clustered point-level geometric features with the node-level semantic feature via our designed feature fusion module. Partial matching is enabled by using a learnable method to select the top-k similar node pairs. Subsequent downstream tasks such as point cloud registration are achieved by running a pre-trained registration network within the matched regions. We further propose a point-matching rescoring method, that uses the node-wise alignment of the 3D scene graph to reweight the matching candidates from a pre-trained point cloud registration method. It reduces the false point correspondences estimated especially in low-overlapping cases. Experiments show that our method improves the alignment accuracy by 10~20% in low-overlap and random transformation scenarios and outperforms the existing work in multiple downstream tasks.

Key facts. Cited by 10 papers in the map (data through 7 Oct 2026). Builds on 40 of its 57 references that are in the map. First posted on arXiv 28 Mar 2024.

Builds on

40 of 57 references.

References · 40

  1. 2017Attention Is All You Needcited by 59,932
  2. 2016PointNet: Deep Learning on Point Sets for 3D Classification and Segmentationcited by 5,223
  3. 2016Visual Genome: Connecting Language and Vision Using Crowdsourced Dense Image Annotationscited by 2,646
  4. 2018Dynamic Graph CNN for Learning on Point Cloudscited by 2,220
  5. 2019SuperGlue: Learning Feature Matching with Graph Neural Networkscited by 1,437
  6. 2019KPConv: Flexible and Deformable Convolution for Point Cloudscited by 1,202
  7. 2021How Attentive are Graph Attention Networks?cited by 818
  8. 20163DMatch: Learning Local Geometric Descriptors from RGB-D Reconstructionscited by 333
  9. 2020PREDATOR: Registration of 3D Point Clouds with Low Overlapcited by 287
  10. 20193D Scene Graph: A Structure for Unified Semantics, 3D Space, and Cameracited by 280
  11. 2019PointNetLK: Robust & Efficient Point Cloud Registration using PointNetcited by 231
  12. 2020Learning 3D Semantic Scene Graphs from 3D Indoor Reconstructionscited by 205
28 more
  1. 2022Geometric Transformer for Fast and Robust Point Cloud Registrationcited by 200
  2. 2019RIO: 3D Object Instance Re-Localization in Changing Indoor Environmentscited by 187
  3. 2022Hydra: A Real-time Spatial Perception System for 3D Scene Graph Construction and Optimizationcited by 175
  4. 2018Learning Latent Permutations with Gumbel-Sinkhorn Networkscited by 172
  5. 2021SceneGraphFusion: Incremental 3D Scene Graph Prediction from RGB-D Sequencescited by 159
  6. 2021CoFiNet: Reliable Coarse-to-fine Correspondences for Robust Point Cloud Registrationcited by 151
  7. 2021Kimera: from SLAM to Spatial Perception with 3D Dynamic Scene Graphscited by 144
  8. 2017Graph-Cut RANSACcited by 133
  9. 2021SOON: Scenario Oriented Object Navigation with Graph-based Explorationcited by 118
  10. 2019Learning Combinatorial Embedding Networks for Deep Graph Matchingcited by 98
  11. 2021Robust Point Cloud Registration Framework Based on Deep Graph Matchingcited by 96
  12. 2018SimGNN: A Neural Network Approach to Fast Graph Similarity Computationcited by 89
  13. 2020Deep Graph Matching Consensuscited by 85
  14. 2021Graph-to-3D: End-to-End Generation and Manipulation of 3D Scenes Using Scene Graphscited by 83
  15. 2021Free-form Description Guided 3D Visual Graph Network for Object Grounding in Point Cloudcited by 81
  16. 2021Optimization of Graph Neural Networks: Implicit Acceleration by Skip Connections and More Depthcited by 55
  17. 2021OMNet: Learning Overlapping Mask for Partial-to-Partial Point Cloud Registrationcited by 50
  18. 2020Neural Subgraph Matchingcited by 47
  19. 2021Learnable Graph Matching: Incorporating Graph Partitioning with Deep Feature Learning for Multiple Object Trackingcited by 47
  20. 2017X-View: Graph-Based Semantic Multi-View Localizationcited by 43
  21. 2023Incremental 3D Semantic Scene Graph Prediction from RGB Sequencescited by 39
  22. 2023SGAligner : 3D Scene Alignment with Scene Graphscited by 30
  23. 2020Retargetable AR: Context-aware Augmented Reality in Indoor Scenes based on 3D Scene Graphcited by 19
  24. 2021Stochastic Iterative Graph Matchingcited by 10
  25. 2023Towards Accurate Loop Closure Detection in Semantic SLAM with 3D Semantic Covisibility Graphscited by 7
  26. 2022Objects Matter: Learning Object Relation Graph for Robust Camera Relocalizationcited by 2
  27. 2022Deep Probabilistic Graph Matchingcited by 1
  28. 2022Leveraging Inlier Correspondences Proportion for Point Cloud Registrationcited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 1; 2025: 4; 2026: 5.

10 papers.

  1. 2025CrossOver: 3D Scene Cross-Modal Alignmentcited by 12
  2. 2025SG-Reg: Generalizable and Efficient Scene Graph Registrationcited by 8
  3. 2025Learning 3D Scene Analogies with Neural Contextual Scene Mapscited by 4
  4. 2024TESGNN: Temporal Equivariant Scene Graph Neural Networks for Efficient and Robust Multi-View 3D Scene Understandingcited by 2
  5. 2025SGAligner++: Cross-Modal Language-Aided 3D Scene Graph Alignmentcited by 2
  6. 2026MA3DSG: Multi-Agent 3D Scene Graph Generation for Large-Scale Indoor Environmentscited by 0
  7. 2026OpenSGA: Efficient 3D Scene Graph Alignment in the Open Worldcited by 0
  8. 2026PROSE: Training-Free Egocentric Scene Registration with Vision-Language Modelscited by 0
  9. 2026From Scene-Centric to Observer-Centric: Modeling Observer-Aware Relations for 3D Scene Graph Generationcited by 0
  10. 2026Graph-GSReg: Leveraging 3D Scene Graphs for Gaussian Splatting Registrationcited by 0