A Depth Camera-Assisted Support Model For Visually Impaired Individuals

Navigating safely and confidently in today’s daily environment poses a significant challenge for visually impaired individuals. Unseen obstacles or unsafe locations can lead to serious accidents. In this paper, we introduce an advanced support system for visually impaired individuals that leverages modern technology to facilitate safer mobility. Our system is built upon point cloud mapping from depth cameras combined with two artificial intelligence models, Yolov8. This integration enables the system to detect obstacles that may cause direct collisions. Another prominent feature of the system is its ability to recognize and notify the user of familiar objects encountered during navigation, with an accuracy of 90.7% and mAP50 is 90.3%. These alerts are delivered to the user promptly, allowing them to react swiftly and avoid potential dangers. The research and experimental results demonstrate that our system can be effectively applied in practical scenarios, representing a significant breakthrough in aiding visually impaired individuals to navigate more safely.