Aplicación móvil con aprendizaje automático para la producción del plátano barraganete en el cantón el Carmen.

Abstract

The purpose of this integrative project is the development of a mobile application with machine learning capacity for the classification of the barraganete banana in the canton of El Carmen, Ecuador. The importance of agriculture in the local economy is undeniable, and bananas are one of the main export products. However, small producers face significant challenges in accurately determining banana ripeness, which can result in considerable econom losses. The research carried out is based on an exhaustive bibliographic analysis and the collection of data through surveys and interviews with producers and agronomist experts. It wa identified that most producers use empirical methods to evaluate banana maturity, which does not always guarantee accurate results. Therefore, the development of an application is proposed that allows farmers to classify bananas as "suitable for export" or "rejection" using artificial intelligence techniques. The development process was carried out using the waterfall methodology, which allowed detailed planning and validation of requirements. A deep learning model was trained using images of different stages of banana maturity, achieving adequate accuracy in the classifications. The results of the evaluation of the application show that it can improve efficiency in banana classification, reduce work time and increase the quality of the final product.