Abstract
Image style transfer is a widely used applied task of automatic redrawing of the source image (content) in the style of another image (setting the target style). Traditional style transfer methods provide only a single stylization result. If the user does not like it, for example, due to artifacts that appear during styling, then he has to choose another style. The paper proposes a modification of the styling algorithm, which gives a variety of styling results with one style, and also improves the average quality of styling by using not only style information from the original style image, but also information from images that have a similar style.
Publisher
Keldysh Institute of Applied Mathematics
Reference11 articles.
1. B. Gooch, A. Gooch, Non-photorealistic rendering, AK Peters/CRC Press, 2001.
2. T. Strothotte, S. Schlechtweg, Non-photorealistic computer graphics: modeling, rendering, and animation, Morgan Kaufmann, 2002.
3. P. Rosin, J. Collomosse, Image and video-based artistic stylisation, volume 42, Springer Science & Business Media, 2012.
4. L. A. Gatys, A. S. Ecker, M. Bethge, A neural algorithm of artistic style, CoRRabs/1508.06576 (2015). URL: http://arxiv.org/abs/1508.06576. arXiv:1508.06576.
5. X. Huang, S. Belongie, Arbitrary style transfer in real-time with adaptive instance normalization, in: Proceedings of the IEEE International Conference on Computer Vision, 2017, pp. 1501–1510.