Clasificador de Fresa Usando Técnicas de Inteligencia Artificial

Abstract

In the development of this research, several artificial intelligence algorithms are implemented for the classification and detection of strawberry maturity level. The images acquired by the sensing system of a conveyor belt were subjected, on the one hand, to Machine Learning and Deep Learning algorithms, oriented to the classification of the fruit into two categories, defined by their maturity characteristics and nutritional deficiencies (visual affectations), and, on the other hand, to Deep Learning algorithms, oriented to the extraction of features to determine their maturity level. A public dataset was created from scratch with the aim of making it available to anyone who needs it. To achieve a satisfactory classification, models were trained using the YOLO architecture, specifically the YOLO architecture, specifically version eight, which allowed us to successfully classify the fruit. fruit classification successfully. Subsequently, tests were carried out on the previously classified images, using techniques such as segmentation were then tested on the previously classified images, using techniques such as thresholded background segmentation and chromatic segmentation. chromatic segmentation. These feature engineering techniques facilitated the detection of fruit maturity level. maturity level of the fruit.