A Fatigue Driving Detection Algorithm Based on Facial Motion Information Entropy

Feng You,Yunbo Gong,Jianzhong Liang,Haiwei Wang,Haiqing Tu

doi:10.1155/2020/8851485

Abstract

Research studies on machine vision-based driver fatigue detection algorithm have improved traffic safety significantly. Generally, many algorithms asses the driving state according to limited video frames, thus resulting in some inaccuracy. We propose a real-time detection algorithm involved in information entropy. Particularly, this algorithm relies on the analysis of sufficient consecutive video frames. First, we introduce an improved YOLOv3-tiny convolutional neural network to capture the facial regions under complex driving conditions, eliminating the inaccuracy and affections caused by artificial feature extraction. Second, we construct a geometric area called Face Feature Triangle (FFT) based on the application of the Dlib toolkit as well as the landmarks and the coordinates of the facial regions; then we create a Face Feature Vector (FFV), which contains all the information of the area and centroid of each FFT. We use FFV as an indicator to determine whether the driver is in fatigue state. Finally, we design a sliding window to get the facial information entropy. Comparative experiments show that our algorithm performs better than the current ones on both accuracy and real-time performance. In simulated driving applications, the proposed algorithm detects the fatigue state at a speed of over 20 fps with an accuracy of 94.32%.

Highlights

Every year, road traffic accidents cause severe damage to human health
As above literature studies discussed, results of the driving fatigue detection have defects of high intrusion, low robustness, and low reliability. erefore, we propose a fatigue driving detection algorithm based on facial motion information entropy. e innovations are as follows: (i) We design a driver’s face detection architecture based on the improved YOLOv3-tiny convolutional neural network and train the network with the open-source data set WIDER FACE [16]
(iii) To get rid of the interference that originated from the size differences between every Face Feature Triangle (FFT), we introduce the face projection datum plane and apply the projection principle to extract the motion feature points of the face. en, based on the motion feature points, we propose the facial motion information entropy, which quantitatively characterizes the chaotic degree of the motion feature points of the face

Summary

Introduction

Road traffic accidents cause severe damage to human health. The Chinese Road Traffic Safety Law stipulates that “Drivers are not allowed to drive continuously for more than 4 hours, and the rest period between every two long-duration driving should be no less than 20 minutes” [3]. In Europe, the law requires that “Drivers should stop and rest for every 4.5 hours of continuous driving, and the rest period should be no less than 20 minutes” [3]. In the United States, the law provision is that “ e cumulative maximum daily driving time must not exceed 11 hours, and the continuous daily rest time must not be less than 10 hours” [4]. It is subjective to determine whether the driver is in fatigue state or not without sufficient quantified indexes and reliable data analysis

Methods

Results

Conclusion