The study hypothesised that a markerless motion capture system can provide kinematic data comparable to a traditional marker-based system for riders mounted on a horse. The objective was to assess the markerless system’s accuracy by directly comparing joint and segment angle measurements taken during walking and trotting with those obtained from a marker-based system. Ten healthy adult participants performed five dynamic trials during walking and trotting. A twelve-camera marker-based system and eight-camera 2D video-based system were synchronised. Three-dimensional hip, knee, shoulder and elbow joint angles, and the global trunk and pelvis angle were computed for comparison between the two systems. To assess the error between systems, the root mean square difference (RMSD) was averaged across each gait cycle and statistical parametric mapping (SPM) paired t-tests were applied. The sagittal trunk angle had the lowest RMSD of 2.0° and elbow rotation had the highest RMSD of 19°, with the same values for walking and trotting. SPM indicated increased hip flexion (0–100 %, p < 0.001) and elbow flexion (24–47 %, p = 0.03; 63–100 %, p < 0.001) in the walking gait cycle for the markerless system. A lack of joint range of motion and obscured medial limbs during walking whilst mounted on horses may cause increased offsets for markerless data in equestrian riders. No significant differences were found for the transverse plane, yet there tended to be increased RMSD. This lack of consistency suggests results from the transverse plane in equestrian riders should be interpreted with caution. Study findings indicate that markerless technology has the potential to be a suitable alternative to marker-based systems for assessment of equestrian riders, dependent on the segment/joint angle of interest and the level of acceptable error. These results indicate that markerless systems can effectively be utilised for rider biofeedback, though their application may be limited for specific joint analyses.

Equestrian

Rider

Kinematics

Deep Learning

Markerless Motion Capture

© 2025 The Author(s). Published by Elsevier Ltd.