Automatic Classification of Plutonic Rocks with Machine Learning Applied to Extracted Shades and Colors on iOS Devices

Abstract

Light/dark shades and color are some properties used for the classification of plutonic rocks but are difficult to measure because they depend on the experience of the observer. Moreover, the classification of plutonic rocks using various instrumental techniques tends to be expensive and time-consuming. To address this situation, we extracted dominant shades and colors from 283 plutonic rock images in RGB and CIELAB formats to train several machine learning models. The best model was deployed on an iOS application that identifies four classes of plutonic rocks from darkest to lightest: gabbro, diorite, granodiorite, and granite. The best results were for the K-Nearest Neighbors model using CIELAB dominant colors data with accuracy, precision, recall, and F-score of 93%.