Semantic-aware neural style transfer
作者：
Highlights：
• We facilitate the segmentation technique to identify the semantics of source images using a domain adaptation technique.
• We propose to utilize word embedding, which guarantees to find semantic matches of the target from the source.
• From the experimental results, we show that the semantic mismatch of neural style transfer is successfully resolved.
• A user study highlights the importance of image semantics for style transfer.
• Our method can be extended to various other style transfer algorithms.
摘要
•We facilitate the segmentation technique to identify the semantics of source images using a domain adaptation technique.•We propose to utilize word embedding, which guarantees to find semantic matches of the target from the source.•From the experimental results, we show that the semantic mismatch of neural style transfer is successfully resolved.•A user study highlights the importance of image semantics for style transfer.•Our method can be extended to various other style transfer algorithms.
论文关键词：Semantic mismatch,Neural style transfer,Segmentation,Domain adaptation,Word embedding
论文评审过程：Received 18 January 2019, Revised 1 April 2019, Accepted 4 April 2019, Available online 11 April 2019, Version of Record 14 May 2019.
论文官网地址：https://doi.org/10.1016/j.imavis.2019.04.001
