The correction model could improve agreement with motion capture by an average of 0.082 on a reliability scale, compared with uncorrected video analysis.The correction model could improve agreement with motion capture by an average of 0.082 on a reliability scale, compared with uncorrected video analysis.