Computer Vision-Based System to Detect Effects of Aromatherapy During High School Classes via Analysis of Movement Kinematics

Abstract

We present non-intrusive visual observation and estimation of movement parameters using RGB data for detecting the effect of olfactory stimulation (essential oils) on movement patterns of high school students during the lessons. In particular, we examine the effect of exposure to aromatherapy has on students’ movement kinetics of upper-body: velocity, acceleration, jerk and energy. The Lavender essential oil was used because of antiseptic, antimicrobial, anti-inflammatory and calming properties that may be used for treating anxiety, insomnia and depression [8, 11, 12]. Two classes were studied, as control and experimental group during two days with week of pause in between. First group had both days without aromatherapy, instead the second - two settings without and with aromatherapy for separate days. For post processing of the recorded data we use OpenPose [7] for estimation of position of joints, Matlab for processing positional data and tracking of the subjects, EyesWeb XMI for the extraction of movement features at a small time scale. Data showed significant differences in velocity, acceleration and jerk for left shoulder and elbow joints of experimental group in comparison between aroma and no aroma settings with Mann-Whitney U test at p < .05. In conclusion, this is an ongoing study shows the possibility of using movement qualities, such as kinematic movement features, extracted ecologically using non-invasive equipment, as a method to measure change of movement behavior, in the cases when no other type of data capture is possible. Future studies will involve further experiments and wider collection of movement features with higher level notations as fluidity, smoothness, rigidity of the movements.