Styleaugment: Learning texture de-biased representations by style augmentation without pre-defined texturesJan 1, 2021ยทSanghyuk ChunSong Parkยท 0 min read PDF CiteTypeJournal articlePublicationarXiv preprint arXiv:2108.10549Last updated on Jan 1, 2021 AuthorsSong ParkResearch Scientist ← Multiple Heads are Better than One: Few-shot Font Generation with Multiple Localized Experts Jan 1, 2021Semantic-aware neural style transfer Jan 1, 2019 →