Visible to the public Separating Style and Content for Generalized Style Transfer

TitleSeparating Style and Content for Generalized Style Transfer
Publication TypeConference Paper
Year of Publication2018
AuthorsZhang, Y., Zhang, Y., Cai, W.
Conference Name2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition
Date Publishedjun
Keywordsbilinear model, Chinese Typeface transfer problem, content encoder, content reference images, convolution, decoder, Decoding, feature extraction, Gallium nitride, generalized style transfer network, Generators, Image coding, image representation, learned model, learning (artificial intelligence), Mixers, multitask learning scenario, neural nets, neural style transfer, Predictive Metrics, pubcrawl, Resiliency, Scalability, Silicon, style encoder, style reference images, Training

Neural style transfer has drawn broad attention in recent years. However, most existing methods aim to explicitly model the transformation between different styles, and the learned model is thus not generalizable to new styles. We here attempt to separate the representations for styles and contents, and propose a generalized style transfer network consisting of style encoder, content encoder, mixer and decoder. The style encoder and content encoder are used to extract the style and content factors from the style reference images and content reference images, respectively. The mixer employs a bilinear model to integrate the above two factors and finally feeds it into a decoder to generate images with target style and content. To separate the style features and content features, we leverage the conditional dependence of styles and contents given an image. During training, the encoder network learns to extract styles and contents from two sets of reference images in limited size, one with shared style and the other with shared content. This learning framework allows simultaneous style transfer among multiple styles and can be deemed as a special 'multi-task' learning scenario. The encoders are expected to capture the underlying features for different styles and contents which is generalizable to new styles and contents. For validation, we applied the proposed algorithm to the Chinese Typeface transfer problem. Extensive experiment results on character generation have demonstrated the effectiveness and robustness of our method.

Citation Keyzhang_separating_2018