Dung Beetle Optimization Based Residual Network Model for Student English Classroom Learning Attention Analysis

For effective teaching, monitoring student engagement and recognizing their attention in class become necessary. Artificial intelligence (AI) based techniques are used for recognizing students’ attention to English classroom learning. As a result, to enhance their interactions, studentattentions, and customization, educational institutions are incorporating the new Machine Learning (ML) technology into their current classroom systems. But even with these tools, it is challenging for teachers to estimate their student’s interest and attentions in English classroom learning. In order to track students’ attention to English classroom learning, this research proposed an advanced real-time vision-based system using Dung Beetle Optimization (DBO) based Residual Network18 (ResNet18) technique. The Beijing Normal University’s Learning Affect Database (BNU-LAD) is used which contains various emotion images. The contrast of an image is enhanced using an image processing technique termed histogram equalization. Then the DBO-ResNet18 automatically extracts the features and chooses the best features for improved students’ attention to English classroom learning recognition. When compared to traditional Convolutional Neural Network (CNN), CNN with Attention Mechanism (CNN-AM), and You Only Look Once version 5 (YOLOv5), the proposed DBO-ResNet18 obtains higher classification accuracy of 99.92%.