Machine Learning to Detect UC Severity

Uses Turi Create Image Classification to categorize the severity of ulcerative colitis from endoscopic images.

Disclaimer

This is for educational uses only!

Installation

Set up a virtual environment, install Turi Create, then navigate to the folder and run

  jupyter notebook

There are 3 notebooks:

1. Dataset To Json - converts raw dataset (link in notebook) into a format Turi can work with.
2. Json To Sframe - loads the dataset into an SFrame. This allows you to delete the original dataset images, since the SFrame contains all images and training data.
3. UC Detector - Trains the model and evaluates performance.

License

MIT