Viet-Anh on Software Logo

What is: Revision Network?

SourceDrafting and Revision: Laplacian Pyramid Network for Fast High-Quality Artistic Style Transfer
Year2000
Data SourceCC BY-SA - https://paperswithcode.com

Revision Network is a style transfer module that aims to revise the rough stylized image via generating residual details image rcsr_{c s}, while the final stylized image is generated by combining r_csr\_{c s} and rough stylized image xˉ_cs\bar{x}\_{c s}. This procedure ensures that the distribution of global style pattern in xˉ_cs\bar{x}\_{c s} is properly kept. Meanwhile, learning to revise local style patterns with residual details image is easier for the Revision Network.

As shown in the Figure, the Revision Network is designed as a simple yet effective encoder-decoder architecture, with only one down-sampling and one up-sampling layer. Further, a patch discriminator is used to help Revision Network to capture fine patch textures under adversarial learning setting. The patch discriminator DD is defined following SinGAN, where DD owns 5 convolution layers and 32 hidden channels. A relatively shallow DD is chosen to (1) avoid overfitting since we only have one style image and (2) control the receptive field to ensure D can only capture local patterns.