Automated Analysis of Teaching Models Based on Artificial Intelligence Detection Algorithms

This paper proposes an automated analysis of teaching models based on artificial intelligence detection algorithms. It aims to efficiently analyze actual classroom teaching models, providing insights that enable appropriate adjustments to teaching strategies, thereby improving the quality of education. We analyze classroom audio data streams using an improved Emphasized Channel Attention, Propagation and Aggregation-Time Delay Neural Network (ECAPA-TDNN) model and analyze video data streams using an enhanced You Only Look Once-v3 (YOLO-v3) model. Subsequently, the obtained information, such as head-raising rates and speaker identification, is automatically processed using an improved S-T analysis method to derive teaching models. Experimental results show that our analytical method, while ensuring that the results closely reflect reality, achieves an analysis speed 2.81 times faster than traditional methods. This demonstrates the advantages of applying automated teaching model analysis in the educational field.