First, get the galaxy data for sdss_images_1000.npy and sdss_labels_1000.npy on astroML's GitHub.
Then copy this data to the root folder of the project.
Execute galaxy_classifier_tf2.py. This will create the weight data for all the galaxies present on the files.
First, get the data from the NASA AWS bucket, the details can be found on this website.
Then copy this data to the root folder of the project.
Execute comet_cnn.py. This will run the model for the regression and give an estimate for the location of a comet in the image.