H-NeXt: The next step towards roto-translation invariant networks
Tomáš Karella (Institute of Information Theory and Automation, Czech Academy of Sciences),* Filip Šroubek (Institute of Information Theory and Automation, Czech Academy of Sciences), Jan Blažek (Institute of Information Theory and Automation, Czech Academy of Sciences), Jan Flusser (UTIA, Czech Academy of Sciences), Václav Košík (UTIA, Czech Academy of Sciences)
The 34th British Machine Vision Conference
Abstract
The widespread popularity of equivariant networks underscores the significance of parameter efficient models and effective use of training data. At a time when robustness to unseen deformations is becoming increasingly important, we present H-NeXt, which bridges the gap between equivariance and invariance. H-NeXt is a parameter-efficient roto-translation invariant network that is trained without a single augmented image in the training set. Our network comprises three components: an equivariant backbone for learning roto-translation independent features, an invariant pooling layer for discarding roto-translation information, and a classification layer. H-NeXt outperforms the state of the art in classification on unaugmented training sets and augmented test sets of MNIST and CIFAR-10.Video
Citation
@inproceedings{Karella_2023_BMVC,
author = {Tomáš Karella and Filip Šroubek and Jan Blažek and Jan Flusser and Václav Košík},
title = {H-NeXt: The next step towards roto-translation invariant networks},
booktitle = {34th British Machine Vision Conference 2023, {BMVC} 2023, Aberdeen, UK, November 20-24, 2023},
publisher = {BMVA},
year = {2023},
url = {https://papers.bmvc2023.org/0578.pdf}
}
Copyright © 2023
The British Machine Vision Conference is organised by