%0 Conference Proceedings
%T Selective Style Transfer for Text
%A Raul Gomez
%A Ali Furkan Biten
%A Lluis Gomez
%A Jaume Gibert
%A Marçal Rusiñol
%A Dimosthenis Karatzas
%B 15th International Conference on Document Analysis and Recognition
%D 2019
%F Raul Gomez2019
%O DAG; 600.129; 600.135; 601.338; 601.310; 600.121
%O exported from refbase (http://refbase.cvc.uab.es/show.php?record=3265), last updated on Tue, 23 Feb 2021 12:55:41 +0100
%X This paper explores the possibilities of image style transfer applied to text maintaining the original transcriptions. Results on different text domains (scene text, machine printed text and handwritten text) and cross-modal results demonstrate that this is feasible, and open different research lines. Furthermore, two architectures for selective style transfer, which meanstransferring style to only desired image pixels, are proposed. Finally, scene text selective style transfer is evaluated as a data augmentation technique to expand scene text detection datasets, resulting in a boost of text detectors performance. Our implementation of the described models is publicly available.
%K transfer
%K text style transfer
%K data augmentation
%K scene text detection
%U https://ieeexplore.ieee.org/abstract/document/8977953
%U http://refbase.cvc.uab.es/files/GBG2019.pdf
%U http://dx.doi.org/10.1109/ICDAR.2019.00134
%P 805-812