Related Experiment Video
Updated: Jul 17, 2026
06:52
Deep-Learning Based Multi-Joint Synchronous Tracking for Objective Quantification of Hindlimb Locomotor Kinematics in Rats
Published on: April 3, 2026
A Synchronized Dual-System Marker-based and Markerless Motion Capture Dataset for Gait and Fundamental Fitness
Zice Guo1, Chunyu Zhao1, Zhengyuan Huang1
1Institute of Artificial Intelligence in Sports, Capital University of Physical Education and Sports, Beijing, 100191, China.
Scientific Data
|July 15, 2026
Summary
This study introduces a new dataset for validating markerless human motion capture systems against marker-based systems. It provides synchronized data for diverse fitness activities, enabling accuracy and consistency evaluations.
Area of Science:
- Biomechanics and Human Movement Analysis
- Computer Vision and Machine Learning
- Sports Science and Kinesiology
Background:
- Marker-based optical motion capture is the gold standard for human movement kinematics.
- Vision-based markerless systems offer improved usability and accessibility but require precision validation.
- Evaluating markerless systems' accuracy in complex movements is crucial for broader adoption.
Purpose of the Study:
- To create a synchronized dual-system dataset for comparing markerless and marker-based motion capture.
- To enable the evaluation of consistency and accuracy between these two systems.
- To provide a resource for developing and assessing movement analysis and action recognition algorithms.
Main Methods:
- Collected synchronized motion capture data from 21 healthy males performing seven fitness activities.
- Utilized seven 120 Hz infrared cameras (marker-based) and two 30 Hz smartphones (markerless).
- Performed technical validation including cross-correlation for synchronization, and regression/correlation analysis for consistency.
Main Results:
- The dataset includes 393 valid records with marker trajectories (.trc) and 2D keypoints (.csv).
- Technical validation confirmed temporal synchronization between marker-based and markerless systems.
- Analysis assessed the consistency of markerless data for trajectory tracking and action recognition.
Conclusions:
- The presented dataset is a valuable resource for the comparative evaluation of motion capture technologies.
- It facilitates the development and validation of algorithms using both marker-based and smartphone-derived markerless data.
- This work supports advancements in accessible and usable human motion analysis.