Development of a Hardware - Software Complex for Recognizing Objects and Determining the Distance to them Based on a Neural Network Model
Abstract
This scientific article is devoted to the development of a hardware - software complex for blind people, based on artificial intelligence, which recognizes objects, measures the distance to the recognized object and provides voice notification in uzbek language about the class of the recognized object and the distance to it. In this research paper, a fine-tuned Yolov5s neural network model based on a special dataset, as well as a stereo camera system for measuring the distance to the recognized object, were used to develop such a complex for recognizing objects. This paper also proposes a solution to improve the accuracy of measurements of the stereo camera system in such projects. The developed project of the complex was experimentally tested. An analysis of the obtained test results was carried out.