Diseño e implementación de un dispositivo para reconocimiento de objetos como apoyo a personas con discapacidad visual

Abstract

This project presents the design and implementation of a smart portable device for object recognition, focused on providing support to people with visual impairments in home environments. The device is primarily targeted at users who have recently lost their sight and who belong to low- and middle-income sectors. It uses a Rasp- berry Pi as the processing unit and an integrated camera, which allows capturing and interpreting images of the controlled environment using pre-trained artificial intelligence algorithms. Technologies such as OpenCV for image processing and PyTorch for the inference of object detection models are analyzed. The device in- tegrates a speaker that produces audible alerts, a vibrating motor for tactile trans- missions, and physical buttons. Validation of the system on the functional prototype demonstrated a real-time detection rate of 4-5 FPS on the Raspberry Pi 4 and a mean approximate pressure (mAP) of 54.7% for the object classes of interest. The system is capable of identifying basic household objects and providing clear feedback to the user, strengthening their autonomy and security, contributing to social inclusion, and aligning with Sustainable Development Goal 3: good health and well-being.