Physical education teaching design under the STEAM concept using the convolutional neural network.

Area of Science:

• Educational Technology
• Artificial Intelligence in Education
• Sports Science

Background:

• Traditional physical education (PE) methods struggle with complex modern research demands.
• Integrating Science, Technology, Engineering, Arts, and Mathematics (STEAM) is crucial for evolving educational paradigms.
• Deep learning (DL) offers potential solutions for enhancing PE teaching and research.

Purpose of the Study:

• To design an efficient deep learning (DL) model for physical education (PE) teaching within the STEAM framework.
• To evaluate the performance of the proposed CNN-STEAM model against traditional CNN and ResNet models.
• To demonstrate the model's effectiveness in data processing and analysis for PE applications.

Main Methods:

• Development of a novel CNN-STEAM model based on convolutional neural networks (CNN).
• Comparative analysis of CNN-STEAM against standard CNN and Residual Network (ResNet) models.
• Performance evaluation using key metrics: accuracy, recall, F1 score, and response time.

Main Results:

• The CNN-STEAM model demonstrated superior performance across all evaluated indicators compared to traditional CNN and ResNet.
• Significant improvements exceeding 20% in accuracy, recall, and F1 score were observed.
• Reduced response times indicate enhanced efficiency for the CNN-STEAM model.

Conclusions:

• The CNN-STEAM model is an effective and efficient deep learning solution for physical education (PE) teaching.
• This model provides robust technical support for PE researchers and educators.
• The study offers novel insights into applying DL within the PE domain under the STEAM concept.