Nobel Chile Jalapeño sorting using structured laser and neural network classifiers

Abstract

Jalapeño chili is grown extensively in Mexico, as it is one of the main vegetables consumed by the population, having also a high demand for exportation. Chili classification is fundamental before arriving to the processing plants, grocery stores and supermarkets. A CCD camera imaged the product which travelled through the conveyor belt, but it was very slow, so a laser scanning system was used to obtain the chili length in order to sort it by sizes. A brief study of the main chili features was carried out, before training a random backpropagation neural network classifier. It was noted that the best topology required to know only the chili width and length sorting up to five different sizes with accuracies over 94%.