Adaptive Driver Face Feature Fatigue Detection Algorithm Research

Fatigued driving is one of the leading causes traffic accidents, and detecting fatigued effectively critical to improving safety. Given variety individual variability surroundings, drivers’ states weariness, uncertainty key characteristic factors, in this paper, we propose a deep-learning-based study MAX-MIN driver fatigue detection algorithm. First, ShuffleNet V2K16 neural network used for face recognition, which eliminates influence poor environmental adaptability detection; second, combined with Dlib obtain coordinates feature points; finally, values EAR MAR are obtained by comparing first 100 frames images EAR-MAX MAR-MIN. Our proposed method achieves 98.8% precision, 90.2% recall, 94.3% F-Score actual scenario application.