Official Tensorflow implementation of U-GAT-IT: Unsupervised Generative Attentional Networks with Adaptive Layer-Instance Normalization for Image-to-Image Translation (ICLR 2020)
hai,thanks for share this project.
demo run well.
course of gpu memory limit, change image_size from 256 to 128, ngf from 64 to 32.
next picture generated after 200,000 iterations.
there are three questions I want to ask.
How many iterations will it take to get the results on the paper?
When change image_size, how to adjust ngf, n_blocks, how to understand?
Have you tried to add face landmark or face segmentation, will it help to preserve face edge , hair edge? Class Activation Map, face landmark which will be better? how to understand?
hai,thanks for share this project. demo run well. course of gpu memory limit, change image_size from 256 to 128, ngf from 64 to 32. next picture generated after 200,000 iterations.
there are three questions I want to ask.