Desenvolvimento de um modelo para detecção de uso de máscara facial e sua adaptação para um dispositivo embarcado

Abstract

This dissertation explores the development of machine learning models tailored for embedded systems. Within this study, diverse models were trained with the objective of classifying mask usage, and subsequently adapted for integration into an embedded system. Model training re lied on an existing dataset comprising five thousand images categorized into two distinct classes. Experimental results suggest the feasibility of incorporating this model, provided specific hard ware constraints such as limited memory, reduced processing capacity, low power consumption, among other aspects, are carefully considered. This research contributes to the understanding of deploying machine learning in resource-constrained embedded environments, with a focus on real-world applications like mask detection.