Convolutional Neural Network based Facial Expression Recognition for Student Engagement Evaluation in Higher Vocational English Classroom

Ensuring high-quality vocational English classroom teaching is essential for fostering effective learning and skill development. An objective and reliable evaluation framework helps educators refine their teaching strategies and academic performance. Traditional evaluation methods rely on subjective assessments, limiting their accuracy and real-time applicability. To overcome these limitations, this research proposes a Convolutional Neural Network (CNN)-based student facial expression recognition system to objectively assess classroom engagement. The AffectNet dataset, containing a diverse range of annotated facial expressions, acts as the basis for training and testing the model. By leveraging deep learning, the proposed method effectively enhances feature extraction, capturing subtle emotional variations indicative of student engagement. The proposed system attains a remarkable classification accuracy of 98.5%, highlighting its performance in recognizing student emotions during lectures. Key performance indicators, including Precision (97.3%), Recall (98.1%), Specificity (97.8%), and F1-Score (97.6%), were thoroughly analyzed, confirming the model's strength and consistency. Experimental findings demonstrate the model's outstanding effectiveness compared to traditional methods, highlighting its potential for real-world implementation. These findings emphasize the practicality of AI-driven solutions in modern education, enabling data-driven decision-making for enhanced learning outcomes.