A Low-Cost Pathological Gait Detection System in Multi-Kinect Environment

Traditional vision-based systems used for automatic gait pathology detection, associate high-cost. However, with the advent of Microsoft Kinect sensor, researchers tried to model some low-cost gait assessment systems; but they suffer from the device-specific generic constraints. This study attempted to mitigate those pitfalls by introducing a noble multi-Kinect setup for automated gait diagnosis. Ten healthy participants were recruited to simulate pathological gait. Extracted salient features were classified using supervised learning, leading to an overall accuracy of 93%, which outperformed state-of-the-art.