Abstract:
Fatigue driving is one of the major contributors to road accidents. Nowadays, COVID-19 is reaching epidemic proportions, which directly leads to the phenomenon of mask-we...Show MoreMetadata
Abstract:
Fatigue driving is one of the major contributors to road accidents. Nowadays, COVID-19 is reaching epidemic proportions, which directly leads to the phenomenon of mask-wearing becomes ordinary among drivers. Most of the existing fatigue detection systems are unable to effectively determine the factual fatigue status of a driver that wearing a mask. Therefore, we propose a quick-witted fatigue detection system to counteract the obstruction of masks. The system detects faces by means of a pyramidbox-based approach. Then a modified PFLD-based method will predict the facial landmarks, from which the eye aspect ratio (EAR) is calculated. Ultimately, our self-made FDUM dataset was tested by using the evaluation method that combined PERCLOS and a method for blink frequency based on Gaussian distribution. Our system can achieve 97.06% accuracy in determining the fatigue status of the driver under the mask, which represents an excellent recognition rate of the system.
Date of Conference: 20-22 December 2021
Date Added to IEEE Xplore: 30 May 2022
ISBN Information: