A single tablet camera can track select walking parameters such as gait speed, but it cannot fully replace multi-camera motion capture systems, according to a comparative evaluation published by researchers evaluating markerless pose estimation. Testing an iPad against an eight-camera markerless motion capture reference across repeated sessions, investigators observed that while mean measurement discrepancies remained relatively small, systematic proportional bias was present across every evaluated spatiotemporal gait metric.
The investigation enrolled thirty-six healthy young adults between eighteen and twenty-seven years of age, with an average age of 18.7 ± 1.7 years. Participants performed walking trials under three distinct task conditions: usual gait, walking while carrying a bag, and walking while using a mobile phone. To assess validity, the team examined 458 paired trials collected during an initial testing visit using repeated-measures and mixed-effects statistical methods. Test-retest reliability was evaluated across four testing visits spaced two weeks apart, analyzing thirty-three participants who completed the full four-visit protocol.
Within-participant associations between the single-camera and multi-camera systems differed substantially by parameter. Using repeated-measures correlation, the strongest association appeared for gait speed (r_rm = 0.588, p < 0.001), followed by stance duration (r_rm = 0.528, p < 0.001), cycle time (r_rm = 0.515, p < 0.001), and cadence (r_rm = 0.446, p < 0.001). Weaker correlations emerged for stride length (r_rm = 0.328, p < 0.001) and step length calculated as half of stride length (r_rm = 0.327, p < 0.001). Meanwhile, swing duration showed no statistically significant association between the two capture setups (p = 0.599).
The researchers identified significant proportional bias across all seven spatiotemporal variables, demonstrating that discrepancies between the single-camera and multi-camera setups varied with the magnitude of the measurement. Furthermore, specific walking conditions influenced measurement bias for cadence, gait speed, cycle time, step length, and stride length. Independent step length and left-to-right step-length asymmetry were not evaluated within the experimental framework.
Across the longitudinal visits, absolute-agreement intraclass correlation coefficients ranged between 0.499 and 0.862. Gait speed maintained good test-retest reliability across all three walking conditions, while step length and stride length achieved their highest reliability during the phone-use task, and swing duration consistently recorded the lowest reliability. The authors concluded that while single-camera capture offers a feasible, accessible alternative for specific metrics like gait speed, multi-camera motion capture remains necessary when three-dimensional reconstruction or comprehensive spatiotemporal characterization is required.
Subject of Research: Validity and reliability of single-camera versus multi-camera markerless motion capture for gait analysis
Article Title: Comparative evaluation of single and multiple camera-based pose estimation methods for gait spatiotemporal parameter estimation
Article References: Weber, B., Schoenwether, B., Booth, T., McFarland, M., Britton, J., & Eltoukhy, M. (2026). Comparative evaluation of single and multiple camera-based pose estimation methods for gait spatiotemporal parameter estimation. PLOS One, 21(11), e0363369. https://doi.org/10.1371/journal.pone.0363369
Image Credits: AI Generated
DOI: 10.1371/journal.pone.0363369
Keywords: gait analysis, single-camera pose estimation, markerless motion capture, spatiotemporal gait parameters, gait speed, cadence, stride length, step length, stance duration, cycle time, swing duration, test-retest reliability, intraclass correlation coefficient, proportional bias, repeated-measures correlation, walking conditions, biomechanics
Cite Scienmag News
Scienmag. (November 21, 2026). Single-Camera Pose Estimation Measures Gait Speed but Cannot Replace Multi-Camera Systems. https://scienmag.com/single-camera-pose-estimation-measures-gait-speed-but-cannot-replace-multi-camera-systems/
Scienmag. "Single-Camera Pose Estimation Measures Gait Speed but Cannot Replace Multi-Camera Systems." Scienmag, 21 November 2026, https://scienmag.com/single-camera-pose-estimation-measures-gait-speed-but-cannot-replace-multi-camera-systems/. Accessed 21 November 2026.
Scienmag. "Single-Camera Pose Estimation Measures Gait Speed but Cannot Replace Multi-Camera Systems." Scienmag. November 21, 2026. https://scienmag.com/single-camera-pose-estimation-measures-gait-speed-but-cannot-replace-multi-camera-systems/








