Modeling of prejudice eyes by designers

Abstract

Research based on human aesthetic sense has been done many times, but that is the theme of this research. In this study, we conducted an experiment to model the aesthetic sense of designers using convolution neural network (CNN) and gradient boost decision tree (GBDT). Moreover, in the proposed method, accuracy was improved by adding colors and character data extracted from other than CNN as new features. The effectiveness of the proposed method was verified by multiple patterns and confirmed.