A Mobile Application For The Visually Handicapped
Rahime Kir, Atike Iscan, Pınar Kirci · 2021 IEEE 16th International Conference on Computer Sciences and Information Technologies (CSIT) · 2021
In this project, an application has been developed to facilitate the lives of visually impaired people and to offer them a better standard of living. This application aims to recognize the objects in front of the user in real time through the camera and announce these recognized objects to the user with the help of a voice assistant. In this way, visually impaired people will be aware of the obstacles in front of them in advance, and they will encounter less of a possible accident. At the same time, they will be able to easily access information about what the objects around them are. YOLOv2 and SDD algorithms were used as object recognition models in the developed project. YOLO (You Only Look Once) is an object recognition algorithm. It also gives the location information of the objects in the image. SSD (Single Shot MultiBox Detector) is an algorithm that uses a single deep network when detecting objects in the image. However, the tiny version of these models was used to improve performance. In this way, the dimensions of the objects are reduced and space is saved and the application works faster.