Fatigue Detection Using Facial Landmarks

Abstract

Detection of fatigue in human face is crucial for medical and safety purposes. Although it is a simple task for a human observer, for a computer, it is a very challenging problem. That is why various attempts were made to successfully detect fatigue. In this paper, we tried to determine the presence of fatigue in a face using computer vision. Here we proposed a method for fatigue recognition by exploiting the facial landmarks. We used the OpenCV library for image processing, and dlib library for feature extraction. The whole method was tested on the extended Cohn-Kanade dataset and the Psychological Image Collection at Stirling (PICS) dataset where it provided a satisfactory level of accuracy.