GAN implementation using Tensorflow
The Following are software are required to run this code:
- NVIDIA CUDA latest version
- Anaconda version 4.3
After installing Anaconda create an environment with python version 3.5 In the environment install the following pakages:
- keras-gpu
- pillow
- opencv
- matplotlib
The input file should be .mat. This code is designed to read .mat file with a struct name 'acitivity' orginazed as followes:
- name (data type string)
- human_number (data type double)
- data ( matrix 600x140 double)
Before running the code check the following script and adjust cordingly:
- GAN_train.py, adjust output directory
- read.py , adjust input directory
Run GAN_train.py to start the magic.
The code will prodcue 1472 images every 300 epoch.