Validation of markerless motion capture for spatiotemporal gait parameters: A comparison between VICON and OpenCap

To evaluate the agreement of spatiotemporal gait parameters obtained using OpenCap, a smartphone-based markerless motion capture system, with those measured using a marker-based optoelectronic motion capture system (VICON) as the reference standard.

Eleven healthy adults performed simultaneous 10-m walking trials recorded using VICON at 100 Hz and OpenCap with two iPhones at 4 K/60 fps. Gait speed, step length, stride length, stride time, cadence, and step width were calculated. Agreement was assessed using intraclass correlation coefficients (ICC), root mean square error (RMSE), mean absolute error (MAE), paired t-tests, and Bland-Altman analysis.

OpenCap showed high agreement with VICON for spatial gait parameters, particularly gait speed, step length, and stride length, with ICC values of 0.98, 0.94, and 0.96, respectively. The RMSE/MAE values were 0.03/0.03 m/s for gait speed, 0.03/0.02 m for step length, and 0.05/0.05 m for stride length. The 95% limits of agreement for these parameters were generally within or comparable to previously reported minimal detectable change ranges. In contrast, cadence showed greater individual-level variability despite an ICC of 0.82, with RMSE and MAE values of 3.14 and 2.35 steps/min, respectively. Stride time and step width showed ICC values of 0.78 and 0.90, respectively, with RMSE/MAE values of 0.03/0.03 s and 0.01/0.01 m, respectively.

OpenCap demonstrated acceptable measurement agreement for selected spatial gait parameters under the present experimental conditions, particularly gait speed and step length. However, cadence showed larger individual-level variability, and temporal and mediolateral parameters should be interpreted cautiously. Further studies in larger and more diverse populations are needed to confirm the clinical applicability of OpenCap-derived spatiotemporal gait parameters.