Apricot Stone Classification Using Image Analysis and Machine Learning

Apricot stones have high commercial value and can be used for manufacturing functional foods, cosmetic products, active carbon, biodiesel. The optimal processing of the is dependent on cultivar there a need methods to sort among different cultivars (which are often mixed in facilities). This study investigates effectiveness two low-cost colour imaging systems coupled with supervised learning develop classification models determine stones. ‘Bella’, ‘Early Orange’, ‘Harcot’, ‘Skierniewicka Słodka’, ‘Taja’ were used. RGB images acquired using flatbed scanner or digital camera; 2172 image texture features extracted within R, G, B; L, a, b; X, Y, Z; U, V coordinates. most influential determined resulted 103 89 selected camera scanner, respectively. Linear nonlinear classifiers applied including Discriminant Analysis (LDA), Decision Trees (DT), k-Nearest Neighbour (kNN), Support Vector Machines (SVM), Naive Bayes (NB). resulting from achieved an accuracy 100% via either quadratic diagonal LDA kNN classifiers. developed all had up 96.77% SVM classifier. presents novel simple-to-implement at-line (flatbed scanner) online (digital camera) methodologies apricot stone sorting. procedure combining machine may authentication quality evaluation sustainable production.