Automatic detection for students behaviors in a group presentation

This paper suggests a model for automatic detection of student behavior to be used in student presentations. The proposed approach is based on the combined use of computer vision libraries and machine learning algorithms to help and support in student assessment using video content. This paper is a part of a research study focusing on investigating and analysing, human behaviours and finding relations between human behaviours and their personal modalities using pattern recognition techniques. For the purpose of this study a group of specific behaviours expressed by students during group presentations in higher education level, are selected. The study proceeds with the detection of the occurrences of those behaviours and comparative analysis of the model's suggested behavioural patterns against those observed through the manual analysis of observations. Both approaches are based on the same set of video files.