isn350 / e_hir_gan

GAN implementation using Tensorflow

e-hir GAN Tutorial

GAN implementation using Tensorflow
The Following are software are required to run this code:

1. NVIDIA CUDA latest version
2. Anaconda version 4.3

After installing Anaconda create an environment with python version 3.5 In the environment install the following pakages:

1. keras-gpu
2. pillow 
3. opencv 
4. matplotlib 

The input file should be .mat. This code is designed to read .mat file with a struct name 'acitivity' orginazed as followes:

1. name (data type string)
2. human_number (data type double)
3. data ( matrix 600x140 double)

Before running the code check the following script and adjust cordingly:

1. GAN_train.py, adjust output directory
2. read.py , adjust input directory

Run GAN_train.py to start the magic.

The code will prodcue 1472 images every 300 epoch.