Visualization Analysis of Learning Attention Based on Single-image PnP Head Pose Estimation

Learning attention analysis of students is the important indicator of classroom teaching/learning quantitative evaluation. Owing to the fact that the head-mounted eye tracker is expensive and unsuitable to be widely used in the large-scale classroom evaluation under expenditure limitation, in this paper, we uses the PnP(Perspective-nPoint) method to estimate student's head pose for single-image. And then we achieve visualization of learning attention. Experiments demonstrate the following advantages of our method. (1) The method limits the average head-pose estimation errors under 4.88° with Biwi database. (2) This work has implemented student learning attention visualization analyses for three typical learning cases including engagement, attention, and disregard.