Evaluation of an Open Access Markerless Motion Capture System During Gait Analysis

In this study, data from virtual markers of the open-access motion capture platform OpenCap were evaluated to record human gait motion simultaneously with a marker-based capture system, considered the gold standard. Four volunteers were recorded, and the kinematic information from both systems was synchronized and processed to obtain joint angles of the hip, knee, and ankle. The results showed a high correlation between the systems in the sagittal and frontal planes, suggesting similarity in motion recording, with an average mean differences for all volunteers of 5° for the hip and 7° for the knee in the sagittal plane, and 3° for the hip and 7° for the knee in the frontal plane. The most significant challenge of this new system (OpenCap) is its inability to estimate the axial rotation of the segments, thus not allowing a reliable assessment of joint kinematics in the transverse plane. Open-access motion capture systems are promising tools, although it is still necessary to minimize differences between markerless and marker-based systems. The evolution of deep learning applications in biomechanics and kinematic information, with a large amount of data, is expected to improve pose estimation and facilitate the use of markerless kinematic recording systems.