Road traffic safety protection model based on IFast-RCNN algorithm

Abstract

The invention discloses a road traffic safety protection model based on an IFast-RCNN algorithm. The method comprises the steps: training data acquisition: determining corresponding labels by using a driving posture of a driver and a road environment picture outside the vehicle shot by a high-definition camera, and uploading the labels to an ASP database; offline model training: acquiring corresponding training data from an ASP database by an upper computer program so as to train the proposed IFast-RCNN algorithm model until the model is converged; online application of the model: inputting the driving posture of a driver and a road environment picture outside the vehicle, which are shot in real time, into the IFast-RCNN algorithm model trained in the step 2 for judgment by a high-definition camera; 4, optimizing and upgrading of the model. The IFast-RCNN model provided by the invention can well protect the road traffic safety, effectively avoids the occurrence of traffic accidents, and has a good practical application value.

Description

technical field [0001] The invention relates to the field of road traffic safety, in particular to a road traffic safety protection model based on the IFast-RCNN algorithm. Background technique [0002] In recent years, my country's urbanization construction, national economy and technological level have been vigorously developed. The number of cars owned by Chinese residents has increased year by year. With the increasing number of cars, traffic accidents will inevitably occur. In recent years, the highway network has basically been formed, and the traffic monitoring system, which is an important part of highway traffic, has been continuously optimized and improved, basically covering the main road sections, which provides a strong guarantee for reducing the incidence of traffic accidents. In fact, most of all kinds of traffic accidents can be avoided. As long as we can warn the driver before the accident, the accident can be effectively avoided. [0003] From the perspect...

AI Technical Summary

Problems solved by technology

National invention patent "A driving safety reminder method, device and medium based on CAN data" (201910926150.8), this method is based on the division of line space and time, defines threshold judgment rules to automatically generate irregular driving behavior rule thresholds, and uses HMM and The k-means clustering method filters the misjudgment in the irregular driving behavior, statistically analyzes the operating trend of the driver and each line and issues a warning, but the definition of the threshold value of the irregular driving in this method is not self-adaptive , which limits the generalization of the method

Images