Future classroom should cover many intelligent features to improve the learning performance of students. Raising hand is one of the most fundamental interactions between students and lectures in classroom. When an intelligent system can automatically understand which students raise their hands, it becomes possible to design more advanced features that can help the teaching in the future classroom. This paper aims to design a practical system for detecting the gesture of raising hand in a real classroom. Getting a video sequence in a classroom, the foreground parts that only contain human bodies are first segmented. We design, therefore, a shape-like descriptor to represent the human bodies and then employ the support vector machine to recognize the student gestures. In order to more precisely understand the student gesture, the student gestures in this work are classified into three classes: left hand, right hand, and normal which means other gestures except raising left or right hand. This work has been implemented and installed in a university classroom. We performed several experiments for this system and demonstrate the results in this paper to present the efficiency.

Full text: IJEIR_835_Final.pdf