Reforming Warning and Obstacle Detection Assisting Visually Impaired People on mHealth

Abstract

Obstacle detection and warning application is to implement for assisting visually impaired people to reduce a harm causing injury. In detail, we study, plan, analyze, design, and develop the obstacle detection module for preventing the visually impaired people crashing the obstructions and send them the alert message for preventing an accidence. Improving the quality of life making them safe and live without anxiety. Regarding methodology, we select an object detection system using on real-time camera or VDO processing named “You Only Look Once” or YOLO. There are two versions; YOLOv2 and YOLO9000, which have tested in this development. YOLOv2 is higher performance than YOLO9000 in 20:7 frames per second (FPS), and higher accuracy. After that, we integrate the detection and notification modules together. Finally, this system can detect and send the alert message to the visually impaired people immediately before they crash the obstructions.