FITBITE: mobile application for the recommendation of healthy meals with image classification

Abstract

The research we present focuses on the development of a mobile application designed for image classification, using a supervised algorithm. This functionality is essential to provide healthy food recommendations, contributing to the maintenance of optimal health standards in users. For the development of this application, we adopted the Scrum framework, structuring it in four main sprints during an estimated period of eight weeks. Regarding its methodological design, this research is non-experimental-descriptive. We evaluated the independent variable through various dimensions and indicators, which are reflected in the questions of a questionnaire applied to a selected sample of 17 individuals. According to the results obtained, the image classification model implemented in the application has a precision (accuracy) of 86.19%. Furthermore, based on the System Usability Scale questionnaire, the usability of the application was rated at 81.029 out of 100 points, which leads us to conclude that the mobile application achieves a level of satisfaction of type A. It is important to mention that, once the application is officially launched, its performance will be evaluated using the JMeter tool.