Drowsiness detection for online courses

Online learning has become more centric in people’s lives today with COVID-19 acting as a catalyst for the growth of this industry. It is believed that this will bring about an educational revolution in the way knowledge is disseminated and become an integral component of the education system. But one should not forget the challenges it brings along with it. The monotony of staring at our screens for so long crepts in disengagement in the form of drowsiness and results in decrease of attention or vigilance on part of the learners. Therefore, with the power vested in our hands through the capabilities of Machine Learning algorithms, we propose an algorithm to detect drowsiness in an individual measured during the online courses using Eye Aspect Ratio (EAR), Mouth Aspect Ratio (MAR), PERCLOS (Percentage of Eyelid Closure over Pupil). Proposed algorithm produces very good results as detailed in this research. It produces an accuracy of approximately 94% and outperforms various other algorithms on the given dataset. Comparative study has also been done in which we have compared proposed algorithm accuracy to other machine learning algorithms’ accuracy. © Grenze Scientific Society, 2021.