Last modified: 2026-05-30
Abstract
Video-based analysis of sprint motion is often affected by limited frame rates, particularly during high-speed lower-limb movements. In this study, high frame-rate video generation was applied to sprint motion analysis to examine its potential usefulness for lower-limb motion evaluation. Synchronized measurements were conducted using a 240 fps video camera and an optical motion capture system to obtain both video recordings and ground-truth motion data.
Sprint trials were performed under three running speed conditions: low, medium, and high speed. Low frame-rate videos derived from the recorded data were processed using an existing frame interpolation model to generate high frame-rate video sequences. Lower-limb keypoint coordinates were estimated from the generated videos using a pose estimation method, and the estimated motion data were compared with optical motion capture measurements to evaluate the influence of frame interpolation on sprint motion analysis.