Serial Skeletal Detection using a Kalman Filter in Combination with OpenPose

The use of OpenPose for gait analysis is highly anticipated as a cost-effective and easy-to-implement method. However, it can sometimes fail to accurately detect skeletal structures, limiting its ability to perform high-precision analysis. Therefore, in this paper, we propose a real-time correction method by combining OpenPose with the Kalman filter for sequential processing. We demonstrate the effectiveness of this approach by comparing the corrected knee joint angles with those obtained from 3D motion capture data.