Enhancing Real-Time Student Emotion Recognition in Online Classrooms Using LSTM and CNN Hybrid Models”

This study aims to design and implement a hybrid model based on Long Short-Term Memory (LSTM) and Convolutional Neural Networks (CNN) for real-time emotion recognition in online classrooms. The system analyzes students’ facial expressions during online learning, capturing their emotional changes in real time and providing immediate feedback to teachers, enabling dynamic adjustments to the teaching content. The model was trained and tested using real datasets, and it demonstrated high accuracy in recognizing typical emotional states such as happiness, confusion, and frustration. Experimental results show that the LSTM-CNN hybrid model excels in both emotion recognition accuracy and system response time, providing emotion feedback within milliseconds, ensuring real-time functionality. After implementing the system, student engagement and classroom interaction significantly improved, with data showing an overall learning effectiveness increase of approximately 12%. This study provides an effective technological solution for personalized teaching in online classrooms and offers strong support for large-scale educational applications in the future.