부산대학교 도서관

Aiming at the characteristics of online teaching mode, this paper focuses on the relationship between facial expressions and teaching effects, and proposes an automatic facial expression recognition model. Use tensorflow software to verify and test the model in the Jaffe dataset. The experimental results show that after the facial expression recognition model performs 100 iterations, the model network reaches the level of convergence. In the training set, the model recognition rate reached 98.76%; in the test set, the model accuracy rate and F1 value reached 1.The model can correctly identify and distinguish Angry, Disgust, Fear, Happy, Sad, Surprise, Neutral 7 facial expressions. The model proposed can capture the changes in facial expressions of students in online classrooms, to help teachers grasp the learning status of students in real time, promote the improvement of online teaching quality, and have certain application value for online teaching quality evaluation.