GAN Framework of Fitting Preference Distribution of “Kawaii”

Abstract

To study male characters reflecting user's preferences in the generation results is a challenging task. In this work, we propose a “kawaii” image generation framework of male animation character faces, which can adapt to Generative Adversarial Network (GAN) frameworks and fit user’s preferences. We present a dataset for the training of the proposed framework. We execute an evaluation experiment to verify the effectiveness of this framework. As a result, our proposed framework is effective, and the generation results of GAN with this framework are more fitted to the user's preference than before.