Diseño e implementación de un prototipo embebido de reconocimiento de objetos por medio de visión artificial como ayuda a personas con deficiencia visual.

Abstract

The current graduate research project was to design and build an electronic prototype for people with visual impairments as a support at the time of moving in indoor environments. Within the work problem, help is given in the social and inclusive field in the development of their daily life through techno ogy and artificial vision. The construction of the prototype was carried out with elements such as Raspberry Pi 4, Pi Noir camera, speakers, among others. The detection system development is based on the “Haar Cascade” algorithm, and its interface was implemented through the Graphic User Interface (GUI) Cascade Trainer. In the first module, images are captured by Pi Noir camera, which are processed on the Raspberry Pi 4 development board; the second module performs a comparison of images, that is, through the algorithm described, the image recognition is obtained with the information about name, position, object size on screen and distance, to finish the third module captures all the previous information, and it sends to the speech conversion sector and transforms it into an mp3 file. The Modules can communicate with each other in a wired way allowing the prototype to indicate if there is an object near the user; therefore, the system issues a voice message with the parameters described, thus facilitating the decision-making for safe movement. The prototype was evaluated by taking several tests of the different objects, of which an average of 87% of effectiveness was obtained. It is concluded that the device can meet expectations to improve the lifestyle of people with visual impairments.