Research Of Driver Attention Detection Based On Monocular Vision

With the development of science and technology and the improvement of living standards,people have higher and higher requirements for the active safety of cars.The analysis results of traffic accident reports in various regions show that the frequent traffic accidents are mostly caused by drivers' inattention.The relevant research in this paper is under the background that intelligent driving assistance system(ADAS)is widely concerned.In this paper,combining with practical engineering projects,the driver attention detection related to intelligent assisted driving is studied.Based on the monocular vision solution,this paper mainly analyzes the correlation between driver attention and driver fatigue and the direction of the driver's line of sight.By studying the driver fatigue state and the direction of the driver's line of sight respectively,the driver's current attention state can be comprehensively judged.In this paper,driver attention detection based on monocular vision is studied.The main application scenes are passenger car driving scenes,and the main purpose of the research is for commercial purposes.Therefore,no organization or individual in the industry has published relevant research data and relevant algorithms.In this paper,in order to solve this problem,the organizer collected the driver's facial data in the simulated scene,and combined with the classification and labeling software proposed and implemented in this paper,annotated the original data,and finally formed the basic data set of this paper.In this paper,a fatigue detection method based on feature fusion is proposed based on monocular vision.This method verifies the correlation between these features and fatigue by analyzing the Angle features and aspect ratio features of various facial organs of drivers,and adopts the method of multi-feature fusion to achieve higher detection accuracy and stable detection effect.The experimental part of this paper verifies the effectiveness of this method.Secondly,based on monocular vision,this paper proposes a recognition method for the estimation of the driver's line of sight by integrating multiple factors.This method estimates the driver's line of sight by analyzing the driver's head posture and the driver's pupil orientation.In this method,the driver's eye behaviors that can cause the driver's inattention are summarized,and two criteria to judge the driver's attention state by the driver's eye behaviors are obtained.The experimental part of this paper verifies the effectiveness of the two basic criteria proposed in this paper,and the experiment also shows that the criteria proposed in this paper can judge the current driver's attention state more accurately.Finally,based on the monocular vision solution and the above algorithm research,the prototype system of driver attention detection is completed.In this paper,the workflow of each module of the prototype system is introduced in detail,and the detection accuracy of each module is analyzed through experiments.