Research On Variable Lane Switching Method Based On Video Detection Technology

In reality,the switching mode of variable lane is mainly based on timing switching,which can no longer meet the complex and changeable traffic needs of intersections.In recent years,a large number of electronic police systems have been constructed and improved,and new breakthroughs have been made in vehicle detection and tracking technology,providing technical support and data support for dynamic switching of variable lane.This paper is devoted to researching the variable lane switching method based on video detection technology,applying the vehicle detection and tracking algorithm based on deep learning to detect and track the vehicles on the entrance road,on this basis,use the video detection method to collect dynamic traffic data,so as to realize the Dynamic switching of lane changes.First,this paper uses the YOLOv5 + Deep SORT algorithm to detect and track various types of vehicles on the entryway.A local vehicle dataset is constructed,and the vehicle objects in the image are divided into three categories: cars,trucks,and buses.The pre-trained YOLOv5 model is migrated and trained using the local data set for a total of 150 cycles,and the model weights when the training effect is optimal are saved.The effect of the YOLOv5 model is evaluated.The m AP(mean Average Precision)value of the model for three types of vehicle detection is 0.938,and the model accuracy is very high.The trained YOLOv5 model is used in combination with the Deep SORT algorithm to track the vehicle,and then the vehicle tracking instance test is carried out.The test results show that the model has a good tracking effect.Then,this paper proposes a collection method for traffic flow and queue length,and analyzes the accuracy and speed of the algorithm.On the basis of vehicle detection and tracking,video detection is used to collect dynamic traffic data.By setting up virtual coils at the entrance road,the traffic flow in all directions of the entrance road and the queue length data of each lane are collected.Then,an example analysis of dynamic traffic data collection based on video detection technology is carried out.The results show that the accuracy and speed of the traffic flow and queue length collection algorithm are good,and the traffic flow and queue length data of the entrance road can be collected quickly and accurately.Finally,this paper proposes a variable lane switching method and a lane function switching evaluation method based on video detection technology.The variable lane switching method based on video detection technology takes the running duration of the variable lane and the queue length as the switching indicators,and uses the green light time of other turns at the entrance to clear the lane.During the lane clearing process,the variable lane is closed.Lane function switching occurs when the last vehicle leaves the variable lane.The lane function switching evaluation method based on video detection technology takes the unit green light time throughput as the evaluation index,and evaluates the lane function switching effect by comparing and analyzing the change of the unit green light time throughput before and after the lane function switching.