Applying simGAN for generating rainfall maps

This repo tested on Python 3.6 and PyTorch 1.4.

The rest of the libraries are in requirements.txt

The real dataset was downloaded from SMHI website (https://opendata.smhi.se/apidocs/radar/)

• Download to tiff images: python utils/dowload_radar_images.py (-h for parser arguments)
• Preprocess: python utils/cut_radar_images.py (-h for parser arguments)

• Run: python utils/random_gaussian_generation.py (-h for parser arguments)

All configuration parameters to train and evaluate RainGAN are in config.py

• In configuration file set is_train=True and is_eyes=False
• Run: python main.py

• In configuration file set load_from_iter, is_train=False and is_eyes=False
• Run: python main.py