So far we have got the best looking result from using DCGAN without any class labels. We intend to make conditional image generation look like the below random generation.
(Note: The code lacks structure and commenting as of now and has various code that is not being used)
