Learning 3D Semantic Scene Graphs from 3D Indoor Reconstructions (CVPR 2020)




Learning 3D Semantic Scene Graphs from 3D Indoor Reconstructions (CVPR 2020)

Johanna Wald*     Helisa Dhamo*     Nassir Navab     Federico Tombari    

Technical University of Munich    Google
* Authors contributed equally.



In our work we focus on scene graphs, a data structure that organizes the entities of a scene in a graph, where objects are nodes and their relationships modeled as edges. We leverage inference on scene graphs as a way to carry out 3D scene understanding, mapping objects and their relationships. In particular, we propose a learned method that regresses a scene graph from the point cloud of a scene. Our novel architecture is based on PointNet and Graph Convolutional Networks (GCN). In addition, we introduce 3DSSG, a semi-automatically generated dataset, that contains semantically rich scene graphs of 3D scenes. We show the application of our method in a domain-agnostic retrieval task, where graphs serve as an intermediate representation for 3D-3D and 2D-3D matching.


Paper

Conference on Computer Vision and Pattern Recognition (CVPR 2020)
Paper | arXiv
  @inproceedings{3DSSG2020,
    title={Learning 3D Semantic Scene Graphs from 3D Indoor Reconstructions},
    author={Wald, Johanna and Dhamo, Helisa and Navab, Nassir and Tombari, Federico},
    booktitle={Conference on Computer Vision and Pattern Recognition (CVPR)}, 
    year={2020}
  }

Dataset

Download: Our 3D Semantic Graphs dataset 3DSSG is available for download here. Please fill out this form if you want access to the 3D reconstructions of 3RScan. A documentation of the data can be found here

Contact

For questions please contact johanna.wald@tum.de and helisa.dhamo@tum.de. For more information regarding the original 3RScan dataset, please visit waldjohannau.github.io/RIO.