Classification And Early Warning System For Day Driver Unsafe Behavior Detection Based On Intelligent Analysis Of Face Video

With the improvement of people’s living standards and the continuous innovation of social production technology,cars have entered ordinary people’s homes and become an important means of transportation for families,but at the same time,the frequency of traffic accidents has increased year by year.The two types of unsafe driving behaviors,fatigue driving and distracted driving,are important causes of traffic accidents,which has also attracted the attention of various countries and has become an important problem in the field of traffic safety research.This article investigates the methods of fatigue detection and distraction detection both at home and abroad,and then conducts comparative analysis,comprehensive pros and cons,and uses OpenFace2.0 facial behavior analysis tools.The aim is to develop a non-contact,real-time and effective driver unsafe behavior detection and warning system.Firstly the face detection method in the system based on the multi-task convolutional neural network model is used to realize face detection,and then the constrained local neural fields model algorithm is used to locate the driver’s facial landmark points.Secondly the head based on camera calibration and direct least squares implements part pose estimation,and then the intersection of the line where the pupil center and the center of the eyeball and the surface of the eyeball achieves the gaze direction of the eye.Finally according to the characteristics of eye aspect ratio,mouth aspect ratio,head yaw angle,and the fatigue and distraction status of eye gaze down to the lower angle,the driver can be warned of unsafe behavior.This paper verifies the effect of face detection and feature point detection algorithms.Under daylight conditions,reasonable thresholds of eye aspect ratio,mouth aspect ratio,head yaw angle,and eye fixation downward angle were obtained through experiments.The early warning effect was tested according to the time elements of fatigue and distraction.The detection and warning system is implemented on the Visual Studio 2015 development platform,and the graphical interface design of the system is completed in Qt Designer.In order to verify the application effect of the system in the actual environment,a test experiment is carried out in a small car to verify the warning effect.The result shows that the designed system has a good accuracy which can meet the requirements of real-time detection and early warning.