An Obstacle Avoidance System for Visual Impaired People Based on Object Tracking Algorithm and Semantic Segmentation

Many visually impaired people face safety issues when walking outside. Providing a safe method to guide such individuals in complex road situations is a widely discussed challenge. With the advancement of deep learning, we have proposed an obstacle avoidance system to address this problem. In this system, a 360-degree camera is used as an input device to record the surrounding environment of users. Deep learning technologies, specifically ByteTrack and MMSegmentation, are employed to process the video data. As a result, the system provides multi-modal feedback to users, helping them avoid obstacles.