A Synchronized Marker-based and Markerless Motion Capture Dataset

This study establishes a dataset designed to validate the accuracy of markerless motion capture systems. Data were collected from 21 male university students majoring in physical education. For each participant, data were acquired synchronously using two distinct systems: (1) a calibrated, high-precision optical motion capture system (OptiTrack) utilizing seven cameras to record anatomical markers at 120 Hz;  (2) a computer vision-based markerless system using two smartphones to capture motion data at 30 Hz. The protocol encompassed seven fundamental fitness movements—including slow walking, fast walking, running, squats, lunges,HighKnees, and jumping jacks—to ensure diversity in both movement intensity and kinematic structure. To facilitate data reproducibility and further research, we provide the raw optical marker coordinates (in C3D format) serving as the ground truth, alongside the synchronized video data.