GAN for everyone
GAN stands for Generative adversarial network.
Generative adversarial networks (GANs) are a class of artificial intelligence algorithms used in unsupervised machine learning, implemented by a system of two neural networks contesting with each other in a zero-sum game framework. They were introduced by Ian Goodfellow et al. in 2014.[1] This technique can generate photographs that look at least superficially authentic to human observers, having many realistic characteristics (though in tests people can tell real from generated in many cases).[2]from wikipedia
easyGAN is a python module which offers an opportunity to try DCGAN easily.
$ pip install easygan
easyGAN has two features. One is to collect images from the internet.
The other is to run DCGAN with images collected by easygan command.
$ easygan --search "google" --keyword "puppy" --folder "puppies"
--search: search engine(currently easygan only suppoprts google image search. will support flickr soon.)
--keyword: search query You can put anything you want.
If you want to use multiple keywords, you need to input the following.
--keyword "NYC food"
--folder: This will be used as a folder name which easygan saves images.
TIP
Before running DCGAN, probably you need to remove some iamges because still image search results are not super acculate. What you need to do is to remove iamges which are not related to the keyword you put.
This phase will take much time if you don't have GPU.
If you run this command successfully, easygan will create the folder, "results" the place you run easygan.
if your images jpg
$ easygan --folder "puppies" --batch 55 --fileformat jpg
--folder: This must be the folder name which has images
If you collect images by yourself and save them in a specific folder, you need to input the name.
--batch: This is batch size.