Application of multispectral imaging combined with machine learning models to discriminate special and traditional green coffee

Abstract

Non-destructive techniques aided by machine learning models are widely implemented in food analysis. To discriminate between ‘special’ and ‘traditional’ classes of green coffee beans, an advanced multispectral imaging technique based on reflectance and autofluorescence data was employed in combination with four machine learning algorithms (SVM, RF, XGBoost, and CatBoost). Of the four algorithms, SVM showed superior accuracy (0.96) for the test dataset. Analysis using PCA and SVM algorithms showed that autofluorescence data from excitation/emission combination of 405/500 nm contributed most to the discrimination of special green coffee from the traditional class. Fluorophores that can be linked to green fluorescence, namely catechin, caffeine and 4-hydroxybenzoic, synapic and chlorogenic acids, were found to have a considerable influence on the differentiation of specialty and traditional coffees. Analysis based on multispectral autofluorescence imaging combined with SVM models was proven to be a valuable tool for future applications in the food industry for the non-destructive and real-time classification of special and traditional green coffee.