Fully Convolutional Geometric Features

Speed vs. Accuracy Pareto optimal frontier of previous methods and ours.

Abstract

Extracting geometric features from 3D scans or point clouds is the first step in applications such as registration, reconstruction, and tracking. State-of-the-art methods require computing low-level features as input or extracting patch-based features with limited receptive field. In this work, we present fully-convolutional geometric features, computed in a single pass by a 3D fully-convolutional network. We also present new metric learning losses that dramatically improve performance. Fully-convolutional geometric features are compact, capture broad spatial context, and scale to large scenes. We experimentally validate our approach on both indoor and outdoor datasets. Fully-convolutional geometric features achieve state-of-the-art accuracy without requiring prepossessing, are compact (32 dimensions), and are 600 times faster than the most accurate prior method.

Paper

paper

Supplementary Materials

• Github
• Supplementary material
• Visualization of correspondences

  Visualization of 500 randomly subsampled correspondences out of ~5k correspondences.

Bibtex

@inproceedings{choy2019fully,
  title={Fully Convolutional Geometric Features},
  author={Choy, Christopher and Park, Jaesik and Koltun, Vladlen},
  booktitle={Proceedings of the IEEE International Conference on Computer Vision},
  pages={8958--8966},
  year={2019}
}