Smart Spectacles To Detect The Household Objects and Familiar Persons For Visually Impairment People Using Single Shot Detector (SSD)

Abstract

According to recent survey, India is ranked third in world’s blind population. India holds 12 million visually impaired peoples as against the global total of 39 million. Peoples with blindness or visual disorder face various issues in carrying out their day to day routine works like travelling, taking medicines, finding the objects, identifying persons, etc. In recent years various voice assist devices helps a lot to assisting blind or visually impaired persons. But still identifying the objects or obstacles, measuring distance between them and warning the blind peoples through audio are still need to be focused more. To solve above mentioned problems, this research work focused on developing a wearable smart spectacles mounted with camera to capture the live video in real world. Then this video is divided into frames and forwarded to the cloud storage. Images that are received in cloud will be processed to identify the object or human and returns the name of the person from the training data or type of object to the mobile application. Later the mobile application integrated with speech interface will pronounce the result in terms of voice. Single Shot Detector (SSD) is a type of object detection algorithm used to detect the multiple objects from the image. For the experimental purpose we have used two type of dataset that contains humans in CCTV footage and household objects.