iMirror: An Intelligent Gait Recognition Mirror for Rehabilitation of Patients with Parkinson's Disease

This paper proposes a novel mirror-based gait recognition system, designated as iMirror, which adopts an AI edge module and walking images capturing by an imaging camera to perform skeleton and gait recognition functionalities. The experimental results show that iMirror can effectively identify the human step length and step shaking angle when walking on the parallel bar. As a result, iMirror is suitable for replacing the traditional mirror in front of the original rehabilitation parallel bar as a tool for analyzing the gait performance of Parkinson's patients.