Attention Decrease Detection Based on Video Analysis in E-Learning

E-learning takes the advantages of lower cost and higher benefit. It becomes one of the educational research focus through learning behavior analysis to promote deep learning. In order to help learners overcome possible disadvantages in e-learning environment such as prone inattention and delayed response, one video analysis algorithm is designed to detect attention decrease situation, then feedback in time or warn early. The algorithm uses head posture, gaze, eye closure and mouth opening, facial expression features as attention observation attributes. Next machine learning classifiers are applied to code behavior features. Finally the time sequential statistics of behavior features evaluate the attention level and emotional pleasure degree. Experiments show that the algorithm is effective to find out the inattention cases to give desirable feedback. It may be applicable in adaptive learning and human computer interaction fields.