SceneCAD (ECCV 2020)

We digitize 3D scans by jointly estimating scene layout and predicting CAD model alignments

Download Data

Please cite our work when using our dataset :)

Description

We introduce a new richly-annotated real-world scene layout dataset con-sisting of 1151 CAD shells and wireframes on top of the ScanNet RGB-Ddataset, allowing large-scale data-driven training for layout estimation.

For the public dataset, we provide annotations with:

13756 Corners
20509 Edges
8376 Planes
1151 real-world 3D scans

Our annotations are:

Non-Manhattan Rooms
Non-Convex Shells
n-Polygonal Planes

Lightweight Layout Prediction Method

Pytorch code coming soon!

Sample Annotations

Citation

If you use this dataset or code please cite:

@InProceedings{avetisyan2020scenecad,
    title={SceneCAD: Predicting Object Alignments and Layouts in RGB-D Scans},
    author={Armen Avetisyan and Tatiana Khanova and Christopher Choy and Denver Dash and Angela Dai and Matthias Nießner},
    booktitle = {The European Conference on Computer Vision (ECCV)},
    month={August},
    year={2020},
}