Learning of Art Style Using AI and Its Evaluation Based on Psychological Experiments

GANs (Generative adversarial networks) is a new AI technology that can perform deep learning with less training data and has the capability of achieving transformation between two image sets. Using GAN we have carried out comparison several art sets different style. We prepared sets; flower photo set (A), an (B1) Impressionism drawings, abstract paintings (B2), Chinese figurative paintings, (B3), images (B4) created by Naoko Tosa, one authors. Transformation A to each B was using four (B1, B2, B3, B4) obtained. these psychological experiment asking subjects consisting 23 students fill in questionnaires. By analyzing obtained questionnaires, found followings. Abstract drawings are clearly judged be different. Figurative West East were similar. Tosa as similar Western images. These results show could used analysis tool reveal differences genres.