The aim of this project is to develop an indoor activity assistance system for the visually impaired. Different from other traditional way to detect the environment, it mainly detects the environment by using deep learning such as the position change of the furniture, obstacle detection and indoor navigation to help visually impaired to avoid dangerous and improve their ability to live independently.

To implement functions development of this project (walk unimpeded, furniture movement detection, signs guidance) combines camera and deep learning without using other sensors. This is not only reduced the cost, but also no need to bring another device.

This project developed the function for visually impaired to avoid the indoor obstacle, indoor guidance and environment detection.