Determination of Cocoa Bean Quality with Image Processing and Artificial Neural Network

Abstract

The objective of this research was to determine the quality of cocoa beans through their digital images. Samples of cocoa beans were scattered on a bright red paper under a controlled lighting condition. A compact digital camera was used to capture the images. The images were then processed to extract their shape and color parameters. Two ANN structures were then used to develop the relationship between input parameters and the bean’s quality components and the outputs. The first ANN having 35 input nodes, classified the bean into 4 beans sizes: whole beans, broken beans, bean fractions, and skin damaged beans. The prediction accuracies were 84%, 52%, 20%, and 20% respectively for the four classes. The low accuracies were caused by the wide variety of beans shapes. The second ANN structure used 6 color parameters to classify the beans into 3 types: normal fermented beans, non fermented beans, and beans with mold. The prediction accuracies were fairly good: 99%, 98%, 79% respectively for the four classes.