Method for automatically identifying and snapshotting illegal parking of vehicles in traffic video based on deep learning

Abstract

The invention discloses a method for automatically identifying and snapshotting illegal parking of vehicles in a traffic video based on deep learning. A vehicle and license plate detection model is built on the basis of a deep learning convolutional neural network technology; the specific type of monitored vehicles (such as buses, taxis and the like) which appear in the traffic monitoring video are detected and identified in real time; the motion states of the monitored vehicles are detected by using an inter-frame differential method to judge whether the monitored vehicles have illegal parking behaviors or not; in combination with a three-dimensional positioning technology and angle rotation and picture zoom of a ballhead camera, clear violated vehicle pictures and license plate number pictures are shot; and automatic identification of license plate numbers is carried out based on a license plate number identification model built by the deep learning convolutional neural network technology, so that automatic snapshotting of violation video clips and structural extraction of violated vehicle information are finally realized, and a complete evidence chain is provided for law enforcement personnel of traffic illegal parking.

Description

technical field [0001] The invention relates to the field of license plate recognition, in particular to a deep learning-based method for automatic recognition and capture of illegally parked vehicles in traffic videos. Background technique [0002] With the vigorous development of social economy and the increase of urban vehicle ownership year by year, the problems of urban road congestion and traffic violations have been increasing, which have become the focus of social attention. Among them, the illegal parking behavior of vehicles not only worsens the congestion phenomenon, destroys the traffic order, but also easily causes traffic accidents and brings great safety hazards. To this end, the city has set up dense traffic cameras to monitor the illegal parking behavior of vehicles. However, in the face of massive traffic surveillance video recordings, it is necessary to manually browse the video segment by segment to locate the vehicle's illegal behavior and illegal segme...

AI Technical Summary

Problems solved by technology

[0005] In view of this, the purpose of the present invention is to overcome the defects in the prior art and provide a traffic video based on deep learning method for automatic identification and capture of illegal parking vehicles, which can solve the problem of low precision and poor effect of automatic identification and capture of illegal parking vehicles in the prior art. poor technical problem

Images