International E-Conference on Engineering, Technology and Management - ICETM 2020
Author(s) : Chao-Hsiang Hung, Hsin-Yu Chen, Kan-Lin Hsiung
Painting is a popular form of art. Style transfer is the technique of recomposing images in the style of other images. In the past, it takes lots of time for a well-trained artist to redraw an image in a particular style. Recently, inspired by the power of Convolutional Neural Networks (CNNs), Gatys et al. published a frontier paper  on how it was actually possible to reproduce famous painting styles on natural images. The main idea behind their proposed algorithm is to iteratively optimize an image with the objective of matching desired CNN feature distribution, which involves both the artwork's style information and the photo's content information. With the appearance of a given artwork, their proposed algorithm successfully produces fantastic stylized images. The seminal work of Gatys et al. has attracted much attention from both industry and academia. In this brief note, we present some preliminary results on neural artistic style transfer for traditional Chinese painting.