Conformal Model for Predicting Interestingness of Microscopy Images based on Windowed Analysis of Image Features

• Intended for the HASTE Storage Client (https://github.com/HASTE-project/HasteStorageClient)
• Analyzes a window of microscopy images for a well-substream
• Queries the HASTE feature database (MongoDB) to retrieve those features
• Performs conformal prediction based on image features
• Applies specified confidence level to return a binary decision.

Usage:

See example.py

Installation:

$ pip3 install git+ssh://git@github.com/HASTE-project/windowed-conformal-model.git@master

or

git clone https://github.com/HASTE-project/windowed-conformal-model.git
cd windowed-conformal-model
pip3 install -e .

Contributors

Ben Blamey, Håkan Wieslander, Phil Harrison