# Geometric Learning of Canonical Parameterizations of 2D-Curves

## TL;DR

This paper introduces a geometric method to handle symmetries in datasets without data augmentation, using fiber bundle theory to improve classification.

## Contribution

The novel contribution is a geometric framework using principal fiber bundles to mod out symmetries in data, enabling class separation without data augmentation.

## Key findings

- A 2-parameter family of canonical curve parameterizations is introduced, including constant-speed parameterization.
- The method effectively mod out symmetries like translation, rotation, scaling, and reparameterization in object contours.
- The approach demonstrates improved class separation by optimizing the section of the fiber bundle.

## Abstract

Most datasets encountered in computer vision and medical applications present symmetries that should be taken into account in classification tasks. A typical example is the symmetry by rotation and/or scaling in object detection. A common way to build neural networks that learn the symmetries is to use data augmentation. In order to avoid data augmentation and build more sustainable algorithms, we present an alternative method to mod out symmetries based on the notion of section of a principal fiber bundle. This framework allows to use simple metrics on the space of objects in order to measure dissimilarities between orbits of objects under the symmetry group. Moreover, the section used can be optimized to maximize separation of classes. We illustrate this methodology on a dataset of contours of objects for the groups of translations, rotations, scalings and reparameterizations. In particular, we present a 2-parameter family of canonical parameterizations of curves, containing the constant-speed parameterization as a special case, which we believe is interesting in its own right. We hope that this simple application will serve to convey the geometric concepts underlying this method, which have a wide range of possible applications.