Unified Style Transfer
This work addresses the lack of standardized evaluation methods in style transfer research, offering a new validation approach that could improve comparability across algorithms.
The paper tackles the problem of comparing and evaluating style transfer algorithms by proposing the Unified Style Transfer (UST) model, which introduces a generative model for internal style representation to handle domain-based and image-based approaches simultaneously, and presents Statistical Style Analysis for validation based on human art sense and style distributions.
Currently, it is hard to compare and evaluate different style transfer algorithms due to chaotic definitions of style and the absence of agreed objective validation methods in the study of style transfer. In this paper, a novel approach, the Unified Style Transfer (UST) model, is proposed. With the introduction of a generative model for internal style representation, UST can transfer images in two approaches, i.e., Domain-based and Image-based, simultaneously. At the same time, a new philosophy based on the human sense of art and style distributions for evaluating the transfer model is presented and demonstrated, called Statistical Style Analysis. It provides a new path to validate style transfer models' feasibility by validating the general consistency between internal style representation and art facts. Besides, the translation-invariance of AdaIN features is also discussed.