Resources
| Description | Link |
|---|---|
| List of coordinates of galaxies and their known classification used for training, in CSV format. | Click Here |
| List of coordinates of galaxies and their known classification used for testing , in CSV format. | Click Here |
| A catalog of 111,838 galaxies from SDSS with predictions about the presence of bar, in CSV format. | Click Here |
| The Python code that uses the trained network for classifying input galaxies. (See instructions below for running the code.) | Click Here |
| The file containing the network structure for deployment. | Click Here |
| The trained Caffe model. | Click Here |
| The 'labels' file needed for running the classification code. | Click Here |
Instructions for running the code
In order to run the code, you will need to install the following dependencies.
- Python 2.7.x
- Caffe
- PIL (Python Imaging Library)
- Astropy
Once you have installed these dependencies, you should download the code, the Caffe model and the labels file given in the table above. The next step is to make a list of your files and store it in a plain ASCII file, let's call it 'imag_list'. You can then run the following command to use the neural network for classifying your galaxies.
$ python classify.py snapshot_iter_471780.caffemodel deploy.prototxt imag_list.lst --labels labels.txt
Before running the code, please be sure to understand the manner in which the network has been trained and its limitations.