Face Expression Recognition Using CNN to evaluate Student Stress Levels in learning comprehension within the VIAT-MAP Application

Technology has significantly transformed education, shifting from traditional classroom instruction to interactive digital learning environments. This study investigates the correlation between student emotional states and comprehension in a digital learning setting using the VIAT-MAP application and an Online Face Expression Recognition System (OFES) implementing Convolutional Neural Networks (CNN) to assess facial expressions in real time. Results from Low-Performance Students indicate a weak negative correlation (R = -0.2302, p = 0.472054) between sad expressions and post-test scores, suggesting cognitive overload among lower-performing students. In contrast, High-Performance Students demonstrated a weak positive correlation (R = 0.052, p = 0.872491) between happy expressions and post-test scores, indicating a slight tendency for higher post-test scores to be associated with higher happiness levels. A moderate positive correlation was observed (R = 0.5758, p = 0.003237) between sad expressions and post-test scores, indicating active engagement. Similarly, a weak positive correlation (R = 0.4405, p = 0.031213) was found between happy expressions and post-test scores, suggesting that happiness levels also tend to increase as post-test scores increase. High-scoring students showed balanced emotional states, correlating with better comprehension. The study concludes that real-time emotional assessment can provide valuable insights into students' learning processes, with emotional expressions associated with academic performance, particularly happiness and a balanced emotional response. Recommendations include expanding the sample size, conducting In-depth Interviews, and implementing a real-time evaluation feature to enhance understanding of emotional impacts on learning.