Domain Generalization vs Data Augmentation: An Unbiased Perspective

Abstract

In domain generalization the target domain is not known at training time. We show that a style transfer based data augmentation strategy can be implemented easily and outperforms the current state of the art domain generalization methods. Moreover, we observe that those methods, even if combined with the described data augmentation, do not take advantage of it, indicating the need of new generalization solutions.

Cite

Text

Borlino et al. "Domain Generalization vs Data Augmentation: An Unbiased Perspective." European Conference on Computer Vision Workshops, 2020. doi:10.1007/978-3-030-66415-2_50

Markdown

[Borlino et al. "Domain Generalization vs Data Augmentation: An Unbiased Perspective." European Conference on Computer Vision Workshops, 2020.](https://mlanthology.org/eccvw/2020/borlino2020eccvw-domain/) doi:10.1007/978-3-030-66415-2_50

BibTeX

@inproceedings{borlino2020eccvw-domain,
  title     = {{Domain Generalization vs Data Augmentation: An Unbiased Perspective}},
  author    = {Borlino, Francesco Cappio and D'Innocente, Antonio and Tommasi, Tatiana},
  booktitle = {European Conference on Computer Vision Workshops},
  year      = {2020},
  pages     = {726-730},
  doi       = {10.1007/978-3-030-66415-2_50},
  url       = {https://mlanthology.org/eccvw/2020/borlino2020eccvw-domain/}
}