A study on gait analysis based on the pose estimation algorithm RTMPose using knee angle as a metric

Gait analysis can be used to assess a person's fall risk and evaluate the effectiveness of fall prevention training. The knee angle is a crucial metric for assessing gait. This study utilized the markerless lightweight pose estimation algorithm RTMPose to evaluate gait. Firstly, 2D human key point detection was performed based on the video. Then the knee angle change curves during walking and pedaling elliptical machine movement were calculated from the key points for gait analysis. This study analyzed knee angle data from individuals with varying body shapes. The results demonstrated that, compared to OpenPose and YOLOv8, RTMPose can quickly and accurately measure the knee angle and determine the abnormal gait condition. It does not need to wear sensors or rely on special testing environments, which is conducive to objective and effective gait analysis by professional rehabilitation workers.