Astro.py

Asteroid Detection in .fits files using Tensorflow Keras.

How to detect

1. Organize your dataset into three folders

>data
  >detect
    >set_of_fits
  >training
    >Asteroids
    >NotAsteroids
  >validation
    >Asteroids
    >NotAsteroids

2. Set parameters according to your dataset in astro.py

3. Run python astro.py

How it works

I've explained it's working in this YouTube video: https://youtu.be/QZC7bWgOmfo

1. Convert .fits to binary PNG using value of x (experimented). Checkout these images for different values of x

   • x=10 (too noisy)
   • x=12 (perfect)
   • x=64 (data lost)

2. Images of same sets are subtracted from each other, and then merged. This will eliminate static objects, only moving ones will be left visible.
   Example

3. Noises left after 2nd step is further reduced using cv2.
   Example

4. After processing all images, training of our model will start. We're using Tensorflow Keras, VGG16 'imagenet'

5. After training completes, model will start detecting Asteroids from Testing set data/detect

---

Contribution:

This is a basic level project, with enough contributions, we can further scale it to greater heights.