hubble space telescope dcgan

• deep convolutional generative adversarial network, dcgan, for creating images resembling those taken by the hubble space telescope.
• the trained generator takes in 100 normally distributed numbers in the range [0, 1] and makes tensors of dim 128, 128, 3 which represent an rgb png resembling the hubble sample data imgs

unprocessed kaggle dataset: https://www.kaggle.com/datasets/redwankarimsony/top-100-hubble-telescope-images?resource=download

timelapse epochs 30-110

training imgs

training sample: opo0010a.png

some generated imgs

the results get worse lol most likely because of overfitting and bad hyperparameters
sm1 save pikachu! 😭 epoch 65:

epoch 70:

epoch 79:

epoch 106:

usage

1. make sure you have python and jupyter notebook installed
2. git clone 'https://github.com/hashirkz/hubble_telescope_gan' <local directory to save to>
3. run all cells in ./usage.ipynb

notes

• the unprocessed kaggle imgs are in compressed .tif format
• the imgs are rgba 4 color channel imgs represented in numpy as m x n x 4 tensors
• be careful reading the imgs into a np.ndarray as some are very large
• recommend to retrain model to around 60 epochs to get organic images
• currently at 110 epochs model is overfitted / hyperparameters may be off a bit because the images look well pretty bad