Vision based Assistive Cane for Visually Impaired

Abstract

Computer vision is a man-made tool that improvises and enables the traditional application processes to transform into an automated process in industrial use cases. This is generalised for several applications that require unguided workflow in our day-to-day life. Our work mainly focuses on reducing the cost factor of an existing vision-based smart assistive cane. Our cane is integrated with a vision system that is capable of notifying the user in case of any obstacle with audio feedback. The main ideology behind this work is to provide a cost efficient assistive cane for the nominally earning visually impaired. The cane is integrated with raspberry pi that handles the object detection and the voice feedback process. A camera is interfaced with the raspberry pi, where the live video feed of the surroundings is obtained and then transferred to the raspberry pi for navigation. Appropriate sounds are given to the user, depending on the surrounding situation and how the user must act about it.