Integrated Intelligent Assistant for Visually Impaired

Abstract A person with a visual impairment finds it more difficult to recognize objects, avoid obstacles, and navigate independently. Visual impairment makes more difficult for those people to learn about their surroundings. Navigating through spaces becomes intricate, as the individual must rely on other senses and aids like canes or guide dogs to detect obstacles and safely maneuver through surroundings. Recognizing objects becomes a complex endeavor, as visual cues are limited or absent, leading to difficulties in identifying items, faces, or even familiar places. The proposed work introduces a brand-new visual assistance solution by creating a single, intelligent device that empowers visually impaired users through its multifaceted functionalities obtained by combining deep learning and machine learning algorithms. Image processing techniques such You Only Look Once (YOLO), Haar-Cascade algorithm is used for object detection and face recognition with the help of camera and other sensors. A comprehensive case study was conducted, demonstrating the real-time capabilities of YOLOv8 in detecting objects and obstacles on public roads, offering empirical evidence of its effectiveness in assisting visually impaired individuals in navigating urban environments safely and independently. The suggested prototype, in comparison to the white cane, provides the visually impaired with increased accessibility, comfort, and ease of navigation, according to the results.