Related Experiment Video
Updated: Aug 26, 2025
06:52
An Inertial Measurement Unit Based Method to Estimate Hip and Knee Joint Kinematics in Team Sport Athletes on the Field
Published on: May 26, 2020
8.1K
Validity of an artificial intelligence, human pose estimation model for measuring single-leg squat kinematics
Lucas D Haberkamp1, Micah C Garcia1, David M Bazett-Jones1
1College of Health and Human Services, University of Toledo, Toledo, OH, USA.
Journal of Biomechanics
|October 5, 2022
Summary
Two-dimensional (2D) pose estimation is a valid tool for assessing adolescent athletes' single-leg squat kinematics. This method accurately captures sagittal and frontal plane joint motion, offering a practical alternative to 3D motion analysis.
Area of Science:
- Biomechanics
- Sports Science
- Movement Analysis
Background:
- Limited research exists on the validity of 2D pose estimation for kinematic analysis, particularly in adolescent populations.
- Assessing lower extremity biomechanics in young athletes is crucial for injury prevention and performance optimization.
Purpose of the Study:
- To evaluate the validity of 2D pose estimation models for analyzing sagittal and frontal plane kinematics during single-leg squats in adolescent athletes.
- To compare 2D pose estimation with 3D motion analysis and traditional 2D methods.
Main Methods:
- Adolescent athletes performed single-leg squats while 3D kinematic data and 2D video (sagittal and frontal planes) were captured.
- Kinematic waveforms and joint angles at peak knee flexion were compared across 2D pose estimation, 3D motion analysis, and traditional 2D analysis.
- Statistical parametric mapping and Pearson's correlation were used for data analysis.
Main Results:
- Significant waveform differences were found between 2D pose estimation and 3D analysis for sagittal ankle and frontal pelvic motion throughout the squat.
- Moderate-to-strong correlations (r=0.68-0.94) were observed for sagittal plane joint angles between 2D pose estimation and 3D techniques.
- Fair correlations (r=0.53-0.54) were found for frontal plane pelvic and hip angles, while poor correlations were noted for frontal knee angles between 2D pose estimation and 3D analysis.
Conclusions:
- 2D pose estimation demonstrates validity as an alternative to 3D motion analysis for evaluating sagittal plane joint angles during single-leg squats.
- While generally accurate for sagittal plane motion, 2D pose estimation shows limitations in capturing frontal plane knee kinematics compared to 3D analysis.
