A New Approach for Counting and Identification of Students Sentiments in Online Virtual Environments Using Convolutional Neural Networks

In this paper, we discuss the importance of counting students and the identification of their sentiments using convolutional networks, specifically YoloV3 and Deepface. First, we discuss the importance of counting and identifying students’ sentiments in online virtual environments; then, we discuss the related work. Later, we present the proposed methodology; we use a repository of fifty images obtained from online virtual classrooms, then we preprocess the images by transforming them to grayscale and normalizing their values to reduce the complexity of processing, and later, we identify and count the number of students by using Convolutional Neural Networks (Yolo-V3), and then by using DeepFace, we recognize the dominant emotion in each student, and in the group, finally, we obtain performances metrics in both stages. The preliminary results show that the identification and counting of students has a precision of 96.7% with a threshold > 50%. The prevalent sentiments in virtual classrooms are neutral 35%, sad 23%, and fear 17%.