Fig 6.

A) Representative example of stillframes from an OpenPose output video of a participant performing the leg agility (i.e., heel tapping) task (left) and violin plots showing comparisons of individual participant mean heel tapping frequencies measured manually (“actual”, blue) and using our pose estimation workflow (“OpenPose”, red) across all four target frequencies (right). B) Violin plots showing event time errors (i.e., differences between video timestamps of heel taps as measured manually and using pose estimation) for each individual heel tap across all participants. C) Scatter plots showing relationships between the video timestamps of all individual heel taps across all participants as measured manually (x-axis) or using our pose estimation workflow (y-axis) across all four target frequencies.