Research On Traffic Safety Warning System Based On Deep Learning

The research on the vehicle driving safety guarantee method is a hot research direction in the driving field.In order to effectively prevent the occurrence of collision accidents in the driving process and timely send out accurate warning signals,an algorithm model that can detect vehicles quickly and accurately and carry out accurate ranging on vehicles is needed.At present,most vehicle detection models are difficult to meet the requirements of early warning system for detection speed,accuracy and model volume.At the same time,the ranging accuracy of traditional ranging methods cannot meet the actual needs of early warning.Aiming at the above problems,this paper proposes a driving safety warning system based on deep learning,and its main work contents are as follows:(1)Aiming at the problems of YOLOv5 s model with large volume,insufficient detection accuracy and slow convergence speed,this paper proposes a lightweight traffic safety warning system.Firstly,Mobilenet V3-small lightweight network is used to replace the backbone of YOLOv5 s network,which greatly reduces the number of network parameters.The size of the model is reduced and the detection speed is improved.At the same time,aiming at the problem of insufficient detection accuracy,several ECAnet attention mechanism modules are integrated into the Mobilenet V3-small backbone network structure in this paper to improve the detection accuracy of the network.Finally,in order to accelerate the convergence speed of the model and further improve the accuracy of detection,In this paper,the original network loss function CIo U＿Loss is replaced by SIo U＿Loss.The experimental results show that the model size of the improved algorithm is reduced by 49.6%.,the m AP is increased by 1.55%,the reasoning time is decreased by 5ms,and the overall performance of the algorithm is significantly improved.(2)Aiming at the problem that the ranging accuracy of traditional ranging methods is affected by vehicle size,this paper proposes a vehicle ranging model based on the center point of the lower edge of the detection frame.By using the monocular visual camera and the vehicle detection algorithm,the position information of the vehicle in front is obtained,and the coordinates of the center point of the lower edge obtained by the vehicle detection frame.The vehicle ranging model was established by synthesizing the pitch Angle information installed by the camera,which solved the error problem caused by the size of the vehicle.At the same time,the trigonometric function model was constructed to solve the X-axis component problem of the front vehicle relative to the experimental vehicle.Experiments show that the accuracy of the improved ranging model is not affected by the size of the vehicle and can take into account the X-axis component of the position of the front vehicle.Compared with the traditional ranging model,the ranging error of the improved ranging model is reduced by about 1.5%,and the ranging accuracy of the improved ranging method is significantly improved.(3)Designed the GUI graphical interface of the traffic safety early warning system,integrated the model selection function,detection mode selection function,distance prompt function and early warning information prompt function,and the improved traffic safety warning system model is transplanted and deployed to EAB310 H embedded device environment,then the static vehicle experiment and the actual road dynamic vehicle experiment were verified.Experiments show that the traffic safety warning system designed in this paper has better vehicle detection and vehicle ranging functions,and can give early warning signal when the distance is below the safety distance.