Comprehensive Teacher Teaching Posture Evaluation Model: Based on OpenPose Posture Recognition in the Context of Digitalization and Intelligence Integration

Zengyi Yu, Zhou Jin, Qinda Xue, Yingchun Liu

Abstract: In the contemporary landscape, the assessment of teaching postures among educators traditionally relies on manual techniques, such as micro-teaching sessions and peer evaluations. However, these methods are often characterized by their inefficiency. Consequently, there is a pressing demand for automated approaches to evaluate and intervene in teaching postures. In the age of data-driven intelligence, this paper introduces the development of an intelligent teaching posture evaluation model. This model is constructed upon the foundation of the OpenPose posture recognition framework, which transforms unstructured video data into structured, analyzable data. Subsequent analysis of this structured data yields valuable metrics for the assessment of teaching postures. Ultimately, through the application of a fuzzy comprehensive evaluation method, this model enables quantitative assessments of teaching postures for both pre-service and in-service educators. The experimental results demonstrate that the evaluations generated by this model closely align with traditional manual assessments. This finding holds substantial implications for enhancing the efficiency and precision of teacher skill evaluations. Additionally, it offers innovative insights and methodologies for supporting educators in improving their teaching postures.