9th International Conference on ICT for Sustainable Development

Authors - Samuel Nghidengwa Nakale, Fungai Bhunu Shava, Gloria Iyawa
Abstract - Employee fatigue is one of the major risk factors across different industries. Consequences of fatigue include injuries and fatalities which in turn affect employee productivity and results in various economic and social costs. Artificial Intelligence (AI) especially Machine Learning (ML) techniques have been used to develop data driven solutions for fatigue monitoring and detection in various industries. However, the fatigue detection systems proposed in literature were predominantly evaluated on simulated data which may not capture some of the real-world driving conditions like in the mining environment. This study deploys three ML classifiers (Support Vector Classifier, Random Forest Classifier and the K-Nearest Neighbor Classifier) to investigate the performance of ML based fatigue detection systems for drivers in the mining industry. The proposed system is implemented in Python and evaluated on a simulated dataset, the Yawning Detection Dataset (YawDD), and a real-world mining operation dataset. All three models achieved a fatigue prediction accuracy above 90% for both datasets. The major challenge to developing behavioral based fatigue detection systems for real world setting like the mining environment is face detection accuracy which is affected by factors such as low image resolution due to poor and variable lighting conditions, face orientation to camera and proximity of face to the camera. The significant contribution of this study is the use of real-world dataset to evaluate the performance of ML based fatigue detection models.