Skip to content

Image-to-image translation using Conditional Adversarial Networks developed with the aim of translation sketch to portrait.

Notifications You must be signed in to change notification settings

TSienki/cGAN-Sketch-to-portrait

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

11 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

cGAN-Sketch-to-portrait

Image-to-image translation using Conditional Adversarial Networks developed with the aim of translation sketch to portrait. Pix2pix architecture was used.

Training dataset was prepared using AISegment.com - Matting Human Datasets. Additionally images were cropped and resized to 256x256 pixel size. There are only portraits images, so in order to obtain sketch images, Photo-Sketching Image project was used.

Some better and worse examples of working of network are presented below.

Using

To use this neural network, you should run script.py with two arguments. First one is path to sketch file (it was tested with sketch file 256x256, but bigger ones also should be good), which will be used to generate new photo. Second one is path, where new photo will be created. Example of cmd command is presented below.

 python .\script.py C:\\x.png F:\\test.png

Due to github's limitation of file sizes, generator model can be downloaded from drive here. This file should be passed to models directory.

About

Image-to-image translation using Conditional Adversarial Networks developed with the aim of translation sketch to portrait.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published