Fatigue Detection During Online Learning Using Facial Features

Abstract

During pandemic Covid-19, university students continuing their study from home to prevent the virus from spread even more. As good as it sounds that students can study from their comfort of their home, there are students suffering from fatigue due to it. This project is to detect if the students having fatigue by using facial to detect it. This is done by using webcam during online classes or online learning where students usually getting fatigue from prolonged visual workload and cognitive activity. The classifier that being tested in this project are Support Vector Machine (SVM), K-nearest Neighbour (KNN) and Extreme Gradient Boost (XGBoost). Feature such as Eye Aspect Ratio (EAR), Mouth Aspect Ratio (MAR), Pupil Circularity (PUC) and MAR over EAR (MOE) being extracted from facial features and fatigue will be evaluate based on Real-Life Drowsiness Dataset. In experiments the results of training and testing the classifiers, SVM shows higher accuracy and F1 score comparing to KNN and XGBoost. The accuracy that the system achieved while tested with the users are 85% of accurately detecting the non-fatigue class and fatigue class. This result shows that the capability of the system in detecting the classes can be reliable.