Cell phone based road condition recognition

Abstract

Visually impaired people, who have a certain degree of damages on the visual function, cannot identify the road condition and pedestrian traffic light. Visually impaired people may need help from other people or guide dogs to go out, which may cause inconvenience. Therefore, it is necessary to investigate the problem. The purpose of this project is to develop an Android application related to road condition recognition. Therefore, the visually impaired or the elderly can go out on their own without others' help. The application, which is able to detect pedestrian traffic lights, shops, and MTR stations, can provide the color of the traffic lights and the name of the shop through voice output. The computer vision technology used to identify and locate the objects on the images or videos is called object detection. The model trained in this paper is based on TensorFlow Objection Detection API.