Real-time object detection and distance measurement for the patients with visual-spatial agnosia

Abstract

Visual-spatial agnosia is a neurological visual impairment disorder that occurs after brain strokes in elderly people. These patients are unable to recognise objects surrounding them and also they cannot pursue the distance of the object. This creates a huge barrier between the real world and the world they perceive. This study proposes a smart eyewear assistive device that will help patients recognize the object as well as guide the patient toward the object by showing the distance of the object with the help of machine learning algorithms. A stereo vision algorithm is used to estimate the distance along with the MaskRCNN algorithm that detects the objects. Physical data has been recorded to analyze the models' performance with the MaskRCNN algorithm with stereo vision algorithm. This study also proposes future works in this area.