Fig. 6
From: Predicting students’ attention in the classroom from Kinect facial and body features
Results on accuracy of attention prediction. Attention estimation accuracies of seven classifiers are shown for each test subject on datasets: a A (normalized features), b A’ (normalized with post-filtering), c B (normalized and smoothed features), d B’ (with post-filtering), and e C (normalized and smoothed features). Graph f represents distributions (shown as boxes) and average accuracies of each classifier over all test subjects of dataset C