Coupled adversarial variational autoencoder

作者：

Highlights：

• A novel model named CoAdVAE is proposed to generate high quality image pairs.

• A suitable loss function is proposed to jointly optimize the parameters of CoAdVAE.

• It has been verified experimentally that the proposed model is effective and efficient.

摘要

•A novel model named CoAdVAE is proposed to generate high quality image pairs.•A suitable loss function is proposed to jointly optimize the parameters of CoAdVAE.•It has been verified experimentally that the proposed model is effective and efficient.

论文关键词：Image pairs,Adversarial variational autoencoder,Resolution,Adversarial learning,Attribute transformation

论文评审过程：Received 12 March 2021, Revised 11 June 2021, Accepted 13 July 2021, Available online 21 July 2021, Version of Record 28 July 2021.

论文官网地址：https://doi.org/10.1016/j.image.2021.116396