ASSESSING LEARNER’S INTEREST IN ONLINE CLASSROOM WITH FACIAL EMOTION DETECTION BASED ON A DEEP LEARNING MODEL

This research aims to develop a tool that helps to evaluate students' interest in online learning through face-to-face emotional detection with a deep learning model. And to find the effectiveness of the tool, it helps to assess student interests in online lessons through facial emotional sensing with deep learning models. It brings facial information into FER2013's instruction and machine learning test, including Neutral, fearful, surprised, happy, angry, sad and disgusted faces, including 7 emotions, as well as facial twisting and curvature detection through a 5-minute video, or 8,018 video frames. A sample of 16 students that chosen by purpose sampling from the Department of Electrical Technology Education, Faculty of Industrial Education and Technology, King Mongkut's University of Technology Thonburi. The research finds that the tool to help evaluate students' interest in online learning through facial emotion detection has the ability to detect students' faces, which represents 87.5 percent of the total student face, which translates the detection performance to a high level. By classifying each emotion on the face detector, up to three orders, which accounted for 31.89, 26.65 and 13.29 percent of all emotions detected, respectively. For the student's face curving, estimate 77.63 percent of the total number of video frames and student face turns that occurred. estimate 4.04 percent of total video framing.