Evaluation of a Kinect-based fundamental movement skills rating system for primary school-aged children
Refereed conference paper presented and published in conference proceedings
CUHK Authors
Full Text
There are no full text file(s) associated with this record. |
Other information
AbstractPurpose: Traditional process-based approaches to measure fundamental movement skills, such as the Test of Gross Motor Development, is typically time-costly and may be subject to differences in trained assessors. To overcome these deficiencies in the assessment of these skills, we developed a system to capture and rate fundamental movement skill performances in real-time using infrared cameras (i.e., Microsoft Kinect). Specifically, motor skills were assessed based on criteria adapted from the Test of Gross Motor Development (versions 2 and 3). In this study, we examined the validity of scores derived from the system.
Methods: 1,174 students (52% male; age = 9.15 ± 1.63 years) from Grades 1 to 6 participated in this study. 1,888 performances of children, including seven ball skills, were rated by the system. The same performances were videotaped and assessed retrospectively by two well-trained assessors. Comparisons of scores were made at the skill level (i.e., combining all scored criteria within each skill). The percentages of agreement and kappa statistics were calculated, for each skill respectively, to evaluate the validity of scores derived from the system.
Results: Fundamental movement skills scored using the system have strong agreement with expert ratings. Percentage agreements and kappa coefficients ranged between 84% to 94%, and .661. to .859, respectively.
Conclusions: Results of the study suggested the developed system could generate valid scores of children’s fundamental movement skills. The objective and real-time feedback generated could improve instruction and learning in physical education. Nonetheless, the system inherited some limitations of the depth sensors, such as limited sensing width and depth, and the inability to detect external objects (e.g., balls). These challenges may be overcome by applications of other advanced techniques, such as machine learning.
Methods: 1,174 students (52% male; age = 9.15 ± 1.63 years) from Grades 1 to 6 participated in this study. 1,888 performances of children, including seven ball skills, were rated by the system. The same performances were videotaped and assessed retrospectively by two well-trained assessors. Comparisons of scores were made at the skill level (i.e., combining all scored criteria within each skill). The percentages of agreement and kappa statistics were calculated, for each skill respectively, to evaluate the validity of scores derived from the system.
Results: Fundamental movement skills scored using the system have strong agreement with expert ratings. Percentage agreements and kappa coefficients ranged between 84% to 94%, and .661. to .859, respectively.
Conclusions: Results of the study suggested the developed system could generate valid scores of children’s fundamental movement skills. The objective and real-time feedback generated could improve instruction and learning in physical education. Nonetheless, the system inherited some limitations of the depth sensors, such as limited sensing width and depth, and the inability to detect external objects (e.g., balls). These challenges may be overcome by applications of other advanced techniques, such as machine learning.
All Author(s) ListHa AS, Cheng J, Chan C, Jiang G, Ng JYY
Name of ConferenceISBNPA meeting
Start Date of Conference18/05/2022
End Date of Conference21/05/2022
Place of ConferencePhoenix, Arizona
Country/Region of ConferenceUnited States of America
Year2022
Month5
LanguagesEnglish-United States
