An Application to Detect Product Adulteration Using Hyperspectral Images & Machine Learning

Abstract

This project develops a machine-learning application for detecting extra virgin olive oil adulteration from RGB and hyperspectral images. A custom dataset of EVOO samples mixed with sunflower oil was collected and used to train RGB regression, HSI regression, and RGB-to-HSI reconstruction models. Although HSI regression achieved higher accuracy, the RGB model was selected for deployment due to its efficiency and accessibility. The final web application allows users to upload oil images, receive adulteration estimates, and share results through a forum and map.