Diseño e implementación de un sistema de bienvenida con reconocimiento facial en un conjunto habitacional

Abstract

The increase in insecurity in Ecuador and in each area within the country increases the probability of being victims, even of intruders within the home. With this degree work, the design and effective implementation of a facial recognition system was concluded, applied within a housing complex located in the north of Quito, specifically in Tumbaco. This system uses image-oriented recognition technologies, specifically in the area of Computer Vision, a branch of Artificial Intelligence. A system was installed to improve internal security, which maintains a record of recognized and unrecognized faces in a simple database. Through an API, it notifies each user with an image of the entry, along with the time and day of registration. The system also provides a personalized welcome to residents and a generic greeting to those who do not belong to the complex. This work addresses the creation of an extra efficient security mechanism for residents and a personalized experience with the use of Computer Vision tools such as Roboflow and YOLOv8. It was implemented using a portable device connected to a camera wirelessly, performing real-time monitoring of users entering the residential complex. Working prototypes were created with models trained with the tools and, finally, the model with the best individual metrics, greater than 0.85, was implemented in the recognition system. With this, through confidence adjustment, facial recognition of the residents was carried out.