Multi-style image transfer system using conditional cycleGAN. (19th May 2021)

• Record Type: Journal Article
• Authors: Tu, Ching-Ting; Lin, Hwei Jen; Tsia, Yihjia
• Is Part Of: Imaging science journal. Volume 69:Number 1/4(2021)
• Page Start: 1
• Page End: 14
• Publication Date: 2021-05-19
• Subjects: Convolutional neural network (CNN) -- deep learning -- generative adversarial net (GAN) -- conditional GAN (CGAN) -- CycleGAN -- conditional CycleGAN -- PatchGAN -- image style transfer; Imaging systems -- Periodicals; 621.36705
• DOI: 10.1080/13682199.2020.1759977
• Languages: English
• ISSNs: 1368-2199

Abstract

ABSTRACT: This paper aims to extend the capability of Cycle-Consistent Adversarial Network (CycleGAN) by equipping it with a conditional constraint and extend it into a multi-style image transfer system that can transfer images among more than two image domains. The conditional constraint is given in the form of the target style feature map instead of a one-hot vector, and has shown to provide better transfer results. The proposed system offers greater flexibility for users to choose the style for image transfer. Experimental results show that such an architecture is not only feasible but also yields good results. The proposed architecture can be extended to other transformation applications, such as facial expressions transfer, face aging, and synthesis of various features.