This project is the implementation of the Paper "Generative Adversarial Networks recover features in astrophysical images of galaxies beyond the deconvolution limit" on python. It is the python version of https://github.com/SpaceML/GalaxyGAN. This python version doesn't include deconvolution part of the paper.
We provide an EC2 AMI with the following pre-installed packages:
as well as the FITS file we used in the paper(saved in ~/fits_train and ~/fits_test)
AMI Id: ami-96a97f80 . (Can be launched using p2.xlarge instance in GPU compute catagory)
Launch an instance.
Please follow the instruction of Amazon EC2.
Linux or OSX
NVIDIA GPU + CUDA CuDNN (CPU mode and CUDA without CuDNN may work with minimal modification, but untested)
We need the following python packages:
tensorflow, cv2, numpy, scipy, matplotlb, pyfits, and ipython
Clone this repo:
git clone https://github.com/SpaceML/GalaxyGAN_python.git
cd GalaxyGAN_python/The data to download is about 5GB, after unzipping it will become about 16GB. Download this file from Google Drive: https://drive.google.com/open?id=1GCs02NBnr7X3skA04hyuXh6cUMZQLzVe
If the mode equals zero, this is the training data. If the mode equals one, the data is used for testing.
python roou.py --input fitsdata/fits_train --fwhm 1.4 --sig 1.2 --mode 0
python roou.py --input fitsdata/fits_test --fwhm 1.4 --sig 1.2 --mode 1XXX is your local address. On our AMI, you can skip this step due to all these have default values.
If you need, you can modify the constants in the Config.py.
python train.py gpu=1You can appoint which gpu to run the code by changing "gpu=1".
This will start the training process. If you want to load the model which already exists, you can modify the model_path in the config.py.
Before you try to test your model, you should modify the model path in the config.py.
python test.py gpu=1The results can be seen in the folder "test".