Automatic Movement Recognition for Evaluating the Gross Motor Development of Infants
The experimental results of pose estimation and tracking indicate that the ViTPose model provided the best performance for pose recognition. A total of 227 features related to kinematics, motions, and postures were extracted and calculated. A one-way ANOVA analysis revealed 106 significant features that were retained for constructing prediction models. The results show that a random forest model achieved the best performance with an average F1-score of 0.94, a weighted average AUC of 0.98, and an average accuracy of 94%.
Authors
- Hui-Ju Chen (ORCID: https://orcid.org/0000-0001-9625-2306)
- Hung‐Yi Chiou (ORCID: https://orcid.org/0000-0002-4545-9697)
- Yang Yu (ORCID: https://orcid.org/0000-0001-5582-4558)
- Mary Ko (ORCID: https://orcid.org/0000-0002-0871-5356)
- Tun‐Wen Pai (ORCID: https://orcid.org/0000-0002-2869-4169)
- Yea‐Ru Yang (ORCID: https://orcid.org/0000-0003-1548-3511)
- Jui‐Chen Tsai
Institutions
- National Taipei University of Technology (TW)
- National Taiwan Ocean University (TW)
- National Health Research Institutes (TW)
- Mackay Memorial Hospital (TW)
- Mackay Medical University (TW)
- Taipei Medical University (TW)
Publication Details
- Journal
- Children
- Published
- 2025-02-28
- DOI
- https://doi.org/10.3390/children12030310
- Citations
- 1
- Primary Topic
- Infant Development and Preterm Care
- Type
- article
- Field-Weighted Citation Impact
- 3.49
Funders
- National Institute for Health and Care Research