Fatigue Detection using the Adaptive Otsu Algorithm

This paper proposes a new approach in the fatigue detection that we did not find it in other studies. The threshold value for the segmentation of “closed”/“open” eye states was determined with an adaptive Otsu algorithm. This algorithm has been optimized to work in real time, using recurrence formulas. The Face Mesh - MediaPipe algorithm was used to detect the face, respectively the delimitation points of the eye positions. The proposed method successfully detects the fatigue state, for the 20 video sequences analyzed, from different people.