Radial base neural network for the detection of banana maturation stages: perceptron multilayer network comparison

Abstract

Agriculture is one of the pillars of human existence since it allows for the obtention of food as well as other products for food production processes. In this regard, there are some crops, such as climactic fruits, that face difficulties especially regarding classification of their maturation stages at the time of harvest, which is the case of bananas, the focus of this work. Therefore, there are some techniques that use artificial neural networks to classify them, such as multilayer networks. Examples of such networks are Perceptron, widely used in several areas, and Radial Base Functional networks (RBF), whose studies are incipient and have little use in agricultural areas. Hence, the objective of the present work was to carry out a comparison between these two neural networks to verify which provides the highest accuracy. In this work it was possible to verify that radial base functional neural networks provide a faster and more efficient categorization for the stages of bananas maturation, because they do not require training and, therefore, have low computational cost, saving more energy, when compared to a Multilayer Perceptron. Therefore, it can be inferred that Radial Base Functional Artificial Neural Networks (RBF ANN) can be widely used in agriculture, enabling the improvement of different cultures and different processes, such as harvesting.

Keywords

• Radial base
• Maturation stages
• Multilayer Perceptron
• Musa acuminata
• Artificial neural networks