Image Style Transferred to Graphical User Interfaces
Résumé
This paper presents an approach for permitting the restyling of an application by automatic analysis of a considered image. The approach relies on the analysis of color distribution from a query image and a color-to-component mapping in order to perform its style transfer to the GUI. No deep learning technique is used, e.g. making the approach applicable without investigating image style learning through training stages. We show how our approach can i) be used to directly and dynamically restyle GUI of applications, ii) be adopted as an enduser functionality towards the restyling of mobile app, desktop app, web app and so on, iii) serve to developers e.g. for facilitating the selection of appropriate reference graphic charter during the design stage of their applications by exploring GUI appearances from a list of considered query images. Experimental results show the efficiency of the approach and its high potential of generalization.
Origine : Fichiers produits par l'(les) auteur(s)