A Blind Aid System based on Jetson TX2 Embedded System and Deep Learning Technique

Blind people always have had many inconveniences in their daily lives because of eye defects, especially walking outdoors has become almost their main challenge and trouble. In order to let blind people easily understand the walking situation ahead to improve the safety and convenience of their walking, this paper proposes a portable blind aid system, which mainly uses the Jetson TX2 embedded system as the development platform and adopts the deep learning technique simultaneously. By utilizing the webcam equipped on the blind person's head and deep learning technique, the front images during his walking are extracted and recognized to judge whether there is his desired target object or not. Then, he can obtain the recognition results via the voice with the use of earphones. Thus, the proposed system can effectively assist the blind people to walk much safely and conveniently by themselves.