Object Detection Using Image Processing for Blind Person

Abstract: This research presents a fully autonomous assistive technology based on artificial intelligence that can distinguish various objects and provides real-time aural cues to the user, improving comprehension for visually impaired people. Multiple photos of items that are extremely useful for the visually impaired person are used to build a deep learning model. The learned model is made more robust by manually annotating and augmenting training photos. A distance-measuring sensor is incorporated in addition to computer vision-based object identification algorithms to enhance the device's comprehension by identifying barriers during navigation. The algorithms used to process images and videos were made to accept inputs from the camera in real-time. Deep Neural Networks were utilized to predict the objects, and Google's well-known Text-To-Speech (GTTS) API module was used to precisely detect and recognize the group or category of objects and locations contained in the anticipated voice message