Class-EyeTention: A Machine Vision Inference Approach of Student Attentiveness’ Detection

Active learning is the key to effective learning in the classroom. Analyzing student participation is very critical in enhancing learning as well as teaching. Attention is the key component in successful learning. However, it is difficult to monitor the attention of individual students in the classroom by using self-reporting. Furthermore, the use of traditional CCTV cameras or video surveillance is intended to reduce the effort involved in manual checking as effectively as possible. In consideration of these difficulties, the use of machine vision to detect student knowledge during the class discussion was suggested. In this research, facial recognition is enabled by an image that has been obtained from student images. The study used three (3) methods to monitor student attention: image detection, webcam detection, and video detection. The aim of this is to continue to process the video captured using traditional CCTV cameras. This will also produce a report on student behavior: attention and not attention. The generated report of the student behavior was based on attention and non-attention level. TensorFlow Lite and Raspberry PI were implemented in this study. In the generated report, students will be informed of one's participation in the class discussion utilizing a time stamp providing details on the behavioral activity conducted by the student. Results show that face recognition results in an accuracy of 96.7% and the detection of attentiveness were also performed.