An Example of using GANs with Flans
Clone or download
Type Name Latest commit message Commit time
Failed to load latest commit information.
example-pretrained-images add example image Nov 16, 2018
pre-trained initial-commit Nov 16, 2018
src/mxnet_gan_flan remove Nov 29, 2018
.gitignore initial-commit Nov 16, 2018
LICENSE initial-commit Nov 16, 2018 Update Nov 16, 2018
project.clj initial-commit Nov 16, 2018

Example GAN Image of Flan


Derived from the MNIST GAN example

This is an example of how to train GANs with your own images. In this example it is flans. To use, you must collect your images and store that in a .rec file. Use this guide to get started

This particular example is for 128x128 images. It takes a bit of memory to run. If you run into troubles try starting smaller like 28x28. There are code comments in the source to help guide you along.


Do lein run and the images generated will be in the results directory. The gout* images are the ones generated, the diff* images are the visualization of the input gradient different fed to the generator

lein run :gpu will run on gpu

You can also work with the repl and explore the pretrained model with explore-pretrained

Or explore your trained model at a specifice epoch with explore

If you are running on AWS you will need to setup X11 for graphics sudo apt install xauth x11-apps

then relogin in ssh -X -i creds ubuntu@yourinstance