YunmeiHuanghi/conditional_gan

How to run the algorithms

There are two python files. Cgan.py use MNIST dataset and the other use tree images I collected.

Method 1

Run the bash file and change hyperparameters.

Method 2

If you don't need to change the hyperparameters, you can run the python file directly.

Datasets explaination

I used two datasets including MNIST and tree canopy images. Run the cgan.py, you will use MNIST dataset. For the tree canopy images, it's my research dataset and not sharable at this moment.

See Reulsts

All the reuslts are saved with tensorboard format. The file will have output in Tensorboard.

Run the following code to load tensorboard and visilusize images and loss figures

%load_ext tensorboard
%tensorboard --logdir run_reuslts_folder

Visulize results in tensorboard

How: use the output folder , then load tensorboard to see images generated by generator and loss curves.
Please make sure you install tensorboard in your running environment. If not, you can directly use Colab to run tensorboard.
You can play with the tensorboard to see the output images at different epoches.

Here is a link with examples to show reuslts by Tensorboard

https://colab.research.google.com/drive/1FcjN8kfBWxYvvu55mj_uP3AN116h6dp8?usp=sharing