This repository contains the codes I am using for my Master's Thesis.
All the datasets and final trained model are uploaded in this Google Drive Link: https://drive.google.com/drive/folders/15Cz5QkND9pC6HosmHSYo_b3DK3SVAEVp?usp=sharing
The main steps involved in this projects is as follows:
-
Generated arbitrary shapes for the transiting object using Bezier curves and save them as image files. I have used one publicly available code to generate the random shapes.
-
Generated transit light curves for the each shape, we used the package EightBitTransit for this simulations.
-
Saved both images and transit light curves as .npy file.
-
We used a 1D CNN machine learning model for mapping transit light curve flux value to the 2D shadow image.
-
After training the model, we tested it on analytical planet light curves from BATMAN package.
Refer this folder to see the training and inference codes: https://github.com/abrahammathews2000/mega/tree/main/ml_training/Google_Colab_Codes
Generating shapes and light curves is little lengthy process (I will try to explain ASAP), till then you could try already saved light curves and shape.