Learner Involvement Recognition in Virtual Classrooms Through Aggregated Neural Classification Techniques

The quick move to online learning has made it much harder to keep track of and keep students interested. Engagement is a crucial element that directly influences learning outcomes; however, conventional methods like manual observation or self-reported feedback are inherently subjective and inefficient. This research suggests an automated system for detecting student engagement through a bagging ensemble deep learning methodology. The system combines different models, such as 1D Convolutional Neural Networks (1D CNN), 1D Residual Networks (1D ResNet), and hybrid architectures, to find patterns in video data over time and behavior. The ensemble approach works better than individual models when using the DAiSEE dataset, with an accuracy rate of up to 94.25%. The suggested framework makes things more stable, more flexible, and able to monitor in real time, giving teachers useful information that can help them teach better.