Görme Engelliler İçin Yapay Zeka Destekli Yapay Görme Sistemi

Abstract

Artificial intelligence assisted artificial vision system for blind individuals

The main purpose of this study is to develop an artificial vision system that has a similar effect to visual perception by using receptive areas on the skin of the dorsum based on sensory substitution methods in order to contribute to the daily life activities of visually impaired individuals. The artificial vision system consists of a camera, a portable mini computer (Raspberry Pi) and a matrix consisting of 20 servo motors mounted on the vest as output unit. Python has been used as a programming language for image processing, and image recognition has been further improved with artificial intelligence capabilities. The performance of the device was evaluated in terms of its ability to transfer correctly to the matrix in 3 different tests. In the socks separation test, the transmission of the correct stimulus to the matrix was checked for black and white socks brought in front of the camera. In the pavement walking simulation test, when approached more than 1 meter to the edges, the device was expected to activate the motors on that side of the matrix. In the walking direction discrimination test, it was expected that the person passing 3 meters away in front of the camera would activate the motors in the walking direction. Ten trials were made for each test and the results were evaluated. In all 10 trials of all three tests, the device performed to transfer data to the matrix as expected. With this study, a new artificial vision system, based on sensory substitution and powered by artificial intelligence that have the potential to contribute to the daily lives of the visually impaired, has been developed.