Detecting Exceptional Actions Using Wearable Sensors' Data for Developing Life-Log Database of Visually Impaired People

This paper proposes our approach to develop a system supporting visually impaired people with life-log. Visually impaired people face daily problems because of lack of information about road conditions. To find the uncomfortable and dangerous points, detecting their exceptional actions, such as tumbling down and getting lost, is effective. In the experiments, we used two units of iPod touch as wearable devices, and measured visually impaired people’s walking as acceleration patterns. As a result, several useful actions appear in acceleration patterns.