Online learning has emerged as a transform- -ative trend in modern education due to its flexibility, accessibility, and ability to deliver personalized learning experiences. However, evaluating student engagement in virtual environments remains a significant challenge, as traditional assessment approaches primarily rely on synchronous, face-to-face interaction, which is often limited or absent in online settings. To address this issue, this paper proposes a novel framework for automated student engagement detection using facial emotion recognition techniques. The proposed system integrates a hybrid deep learning architecture that combines Convolutional Neural Networks (CNN), Convolutional Block Attention Module (CBAM), and Temporal Convolutional Networks (TCN) to effectively capture both spatial and temporal characteris- -tics of facial expressions. CNN is utilized to extract robust spatial features from facial images, while CBAM enhances feature representation by emphasizing salient spatial and channel-wise information. Furthermore, TCN models sequential emotional variations across video frames, enabling comprehensive analysis of dynamic engagement patterns. To support the development and evaluation of the model, a specialized dataset was constructed to capture diverse engagement scenarios in realistic online learning environments. Experimental results demonstrate that the proposed system achieves a classification accuracy of 92.3%, outperforming baseline CNN models and ensuring reliable real-time performance. The framework provides actionable insights for educators and establishes a scalable, intelligent solution for enhancing engagement analytics in virtual education systems.