MARWAH AMAN

Key Skills: Sensor Integration, Python, Machine Learning, Prototyping

​

The design has been patented under Marwah Sabrah.

​

A produce ripeness scanner, machine learning model, and Android app that interface to predict the exact ripeness date of any avocado with +/- 1-day accuracy.

This scanner collects spectral data, electrical impedance data, and sound attenuation data when scanning an avocado.

I preprocessed 250 avocado scans using PCA (principal component analysis) and used that data to train 15 different supervised machine learning models from the scikit-learn python library. From there, I narrowed down to the most accurate model and used that for avocado ripeness prediction.

The scanner and machine learning algorithm are all integrated and accessed through an android phone application that we developed called Avocado. This phone application instantly displays ripeness information for new scans, shows batch information for batches of avocados scanned, includes a scan history page of statistics on all of the user’s scanned avocados, and uses a sign in/sign out feature to retain user data.