This repository is not meant be used in production or to product real usable data. It is still incomplete and not perfect. For now it only serves the purpose of being supplementarial materials for my report available here, beware it's in french.
The dcgan.py
file is meant to define a general structure of the GANs used. gan_training.py
is an example of DCGAN training and cgan_training.py
an example of CDCGAN training.
Some data-sets used in my report can be found in the Samples directory. A README.md
file briefly describes each file.
These directories are meant to be used while training GANs with the python files presented above.
Feel to propose any kind of amelioration. From my new expertise, the generative model used might be to complexe and maybe a simplification of it could improve the models accuracy.