Determination of the Ripeness State of Guavas Using an Artificial Neural Network

The determination of the ripeness state of fruits is an essential element in the agriculture research field. This is because the ripeness is related with quality and it can affect the commercialization of the product. In this paper, a classification system of the ripeness state of guavas is proposed. The guavas are classified into three states: green, ripe and overripe. The classification system is based on Artificial Neural Network (ANN) which uses color features as input. The characteristics used in our proposal are extracted from three different color spaces: RGB, CIELab and CIELuv. Specifically, we only use the components R, G, a and u, which gave us the best separability within classes. The system was tested using real images of guavas obtaining 97.44% of accuracy.