TY - GEN
T1 - Research on evaluation algorithm of teacher's teaching enthusiasm based on video
AU - Chen, Yujia
AU - Wang, Chongwen
AU - Jian, Zefeng
N1 - Publisher Copyright:
© 2020 ACM.
PY - 2020/11/20
Y1 - 2020/11/20
N2 - Most of the current research on teachers' teaching enthusiasm is to define teachers' emotions in a qualitative way, but lacks intuitive digital or image representations. Based on the online teaching platform, this research collects various types of teaching videos, and establishes a teacher's teaching emotion data set based on the PAD emotion model, then extracts teacher's teaching features through various algorithms such as sound feature extraction, facial expression recognition, target detection, and pose estimation. Analyze the features of different modalities, and cascade the features based on the contribution of each feature to the result. Finally, use BP neural network to predict teachers' enthusiasm. The study found that the model prediction performance is better after adopting the cascading feature fusion method based on feature contribution. The loss of the P prediction model is 0.0586, the loss of the A prediction model is 0.0517, the loss of the D prediction model is 0.0428, and the average loss is 0.0510.
AB - Most of the current research on teachers' teaching enthusiasm is to define teachers' emotions in a qualitative way, but lacks intuitive digital or image representations. Based on the online teaching platform, this research collects various types of teaching videos, and establishes a teacher's teaching emotion data set based on the PAD emotion model, then extracts teacher's teaching features through various algorithms such as sound feature extraction, facial expression recognition, target detection, and pose estimation. Analyze the features of different modalities, and cascade the features based on the contribution of each feature to the result. Finally, use BP neural network to predict teachers' enthusiasm. The study found that the model prediction performance is better after adopting the cascading feature fusion method based on feature contribution. The loss of the P prediction model is 0.0586, the loss of the A prediction model is 0.0517, the loss of the D prediction model is 0.0428, and the average loss is 0.0510.
KW - BP neural network
KW - Cascading feature fusion
KW - PAD emotion model
KW - Teaching enthusiasm
UR - http://www.scopus.com/inward/record.url?scp=85108168850&partnerID=8YFLogxK
U2 - 10.1145/3449301.3449333
DO - 10.1145/3449301.3449333
M3 - Conference contribution
AN - SCOPUS:85108168850
T3 - ACM International Conference Proceeding Series
SP - 184
EP - 191
BT - 2020 6th International Conference on Robotics and Artificial Intelligence, ICRAI 2020
PB - Association for Computing Machinery
T2 - 6th International Conference on Robotics and Artificial Intelligence, ICRAI 2020
Y2 - 20 November 2020 through 22 November 2020
ER -