Skip to content

ymzayek/asteroidsCNN

Repository files navigation

Asteroids CNN

CNN for detecting asteroid images from the Kilo-Degree Survey (KiDS)

Based on https://github.com/MikkoPontinen/EuclidCNN


Training

File trainSimpleCNN.py can be used to train a model from scratch. From the command line, the following arguments need to be set:

  • --image_path <path-to-image-files>
  • --model_out <path-to-save-model>
  • --plots_out <path-to-save-plots> [optional]

Example:

python3 trainSimpleCNN.py --image_path /data/pg-ds_cit/Projects/Astronomy/AstronomyProject/Images --model_out /data/p301081/astronomy/test --plots_out /data/p301081/astronomy/test

A training example can be found in the notebook trainSimpleCNN.ipynb

The training script also includes testing model performance with a test set.

About

binary classification of astronomical images

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published