Visual Assistance Software for the Visually Impaired

Abstract

The proposed project involves the creation of a visual assistance software for the visually impaired to identify people in their surroundings to avoid potential collisions. Object detection technique using TensorFlow Lite Object Detection API is performed to identify the people in the view of a visually impaired person. Vibrations of different lengths were used to notify the visually impaired about the location of the person in the view. The software is tested in various conditions to verify its working.

Presented in absentia on April 27, 2020 at "Student Research Day" at MacEwan University in Edmonton, Alberta. (Conference cancelled)

Faculty Mentor: Jeffrey Davis

Department: Computer Science