Stylistic mixture of Monet and Chinese ink painting by deep learning

---

Xupu Geng and Tian Li

Xiamen University, China

: J Comput Eng Inf Technol

Abstract

---

Image style transfer is a classical problem in computer graphics and vision. As the palmy development of deep learning in recent years, Generative Adversarial Networks (GAN) and its variations like CycleGAN have been proposed to generate or transform images. Monet and Chinese Ink are two influential art styles in landscape painting. They have some likeness in impressionism but concerning color and depth of focus, they are so different. Here we try to mix the two styles to create a new kind of artwork by CycleGAN. In fact, the proposed method in this paper has many potential applications in artistic creation.

Biography

---

Xupu Geng is Senior Engineer in State Key Laboratory of Marine Environmental Science, Xiamen University. His research interests are in deep learning and its application in image processing.

E-mail: gengxp@xmu.edu.cn