Vision based composite approach for lethargy detection

This paper presents a fatigue detection techniques based on computer vision. Fatigue is detected from face and facial features of driver. Hybrid method is used for face and facial feature detection which not only increase the accuracy of the system but also decrease the processing time. Skin color pixels detection and viola Jones methods is used for face detection and knowledge based division method is used to increase the accuracy of facial feature detection. Also a dynamic threshold value is used for yawning and eyes status detection. The average yawning false alarm rate is 1.5% and closed eyes false alarm rate is 0.13%. Algorithm is developed in MATLAB software. The average accuracy of the system is 97.7%.