Thesis Aplicacion móvil de aprendizaje culinario gamificado - Backend
Loading...
Date
2025-01
Authors
Journal Title
Journal ISSN
Volume Title
Program
Ingeniería Civil Telemática
Departament
Campus
Campus Casa Central Valparaíso
Abstract
El aprendizaje culinario en plataformas digitales suele carecer de retroalimentación práctica y personalización, limitando la motivación y el progreso del usuario. Esta memoria presenta el diseño y desarrollo de la arquitectura técnica de SHEFU, una aplicación móvil gamificada que integra modelos de visión computacional para la evaluación automática de preparaciones culinarias. La solución se construyó sobre una arquitectura modular, utilizando Flutter para el cliente y Firebase como backend serverless (BaaS), gestionando la autenticación, la persistencia de datos y la lógica de negocio. Se destaca la implementación del módulo de Visión Computacional, el cual ejecuta un modelo EfficientNet-Lite0 directamente en el dispositivo (on-device) mediante TensorFlow Lite. Esta estrategia permitió evaluar la presentación de los platos mediante criterios objetivos, con una latencia de inferencia inferior a 3 s, garantizando una experiencia fluida sin dependencia de conectividad constante. Adicionalmente, se desarrolló un sistema de gamificación integral que gestiona rankings, moneda virtual y una tienda de accesorios. La validación del sistema, realizada con usuarios finales, arrojó un puntaje de usabilidad SUS de 85.0 (grado Excelente) y un NPS de +50, confirmando una alta aceptación y lealtad. Los resultados demuestran que la arquitectura propuesta es robusta y escalable, soportando eficientemente la integración de tecnologías avanzadas en un entorno móvil para mejorar la experiencia de aprendizaje.
Culinary learning on digital platforms often lacks practical feedback and personalization, limiting user motivation and progress. This thesis presents the design and development of the technical architecture for SHEFU, a gamified mobile application that integrates artificial intelligence for the automatic evaluation of culinary preparations. The solution was built upon a modular architecture using Flutter for the client and Firebase as a serverless backend (BaaS), handling authentication, data persistence, and business logic. A key contribution is the implemen tation of the Computer Vision module, which runs an EfficientNet-Lite0 model directly on-device using TensorFlow Lite. This strategy allowed for the evaluation of dish presentation based on objective criteria with an inference latency of less than 3 s, ensuring a fluid experience without reliance on constant connectivity. Additionally, a comprehensive gamification system was developed to manage rankings, virtual currency, and an accessories store. System validation with end-users yielded a SUS score of 85.0 (Excellent grade) and an NPS of +50, confirming high acceptance and loyalty. The results demonstrate that the proposed architecture is robust and scalable, efficiently supporting the integration of advanced technologies in a mobile environment to enhance the learning experience.
Culinary learning on digital platforms often lacks practical feedback and personalization, limiting user motivation and progress. This thesis presents the design and development of the technical architecture for SHEFU, a gamified mobile application that integrates artificial intelligence for the automatic evaluation of culinary preparations. The solution was built upon a modular architecture using Flutter for the client and Firebase as a serverless backend (BaaS), handling authentication, data persistence, and business logic. A key contribution is the implemen tation of the Computer Vision module, which runs an EfficientNet-Lite0 model directly on-device using TensorFlow Lite. This strategy allowed for the evaluation of dish presentation based on objective criteria with an inference latency of less than 3 s, ensuring a fluid experience without reliance on constant connectivity. Additionally, a comprehensive gamification system was developed to manage rankings, virtual currency, and an accessories store. System validation with end-users yielded a SUS score of 85.0 (Excellent grade) and an NPS of +50, confirming high acceptance and loyalty. The results demonstrate that the proposed architecture is robust and scalable, efficiently supporting the integration of advanced technologies in a mobile environment to enhance the learning experience.
Description
Keywords
Flutter, Firebase, TensorFlow Lite, Arquitectura serverless, Visión computacional On-Device, Gamificación
