We recommand to use the notebook (.ipynb) but the Python script (.py) is also provided if more convenient for you.
# GAN & cGAN tutorial.
# How to submit your Work ?
Matheus MACHADO E SILVA
This work must be done individually. The expected output is a repository named gan-cgan on https://gitlab.ec-lyon.fr. It must contain your notebook (or python files) and a README.md file that explains briefly the successive steps of the project. The last commit is due before 11:59 pm on Wednesday, March 29, 2023. Subsequent commits will not be considered.
# GAN
GAN is a generative adversarial network that we use to generate new handwritten digits, after feeding it with pictures from the MNIST database;
To reduce the time of execution we use Google Colab GPU;
The MNIST dataset uses black and white images so to adapt this tutorial to the MNIST database, I changed the number of channels for 1 (in the tutorial was 3 because of RGB data used);
# DCGAN
# cGAN tutorial.
To reduce the time of execution we use Google Colab GPU;
The MNIST dataset uses black and white images so the first step is to change the number of channels for 1 (in the tutorial was 3 because of RGB data used);
cGAN is a conditional generative adversarial network (cGAN) and in this project I trained it to identify if building facades images have generated by a model or are real;
The data used is from the CMP Facade Database;
To generate the facade images, a mask, already included in the database, passes through the generator, which delivers as output a believable image.
