The purpose of the project implementation plan is to generate fake image through gan.
I used BEGAN(boundary equilibrium generative adversarial networks)
There are three sample fake images produced from began.
I also transfer the original Celeba data to high quality image celeba HQ as my input data.
In the paper of Progressive Growing of GANs, it provides a way to get such high-resolution images.
I create this data from celeba-hq-modified
In this task, I tried to produce different resolution image such as 64x64, 80x80, 100x100 and 112x112.