Deep learning and background difference method fused Safe City traffic flow counting method

Abstract

The invention discloses a deep learning and background difference method fused Safe City traffic flow counting method, and aims at overcoming defects of infrared, ground induction coil and supersonicwave detection methods. A background difference method is used to obtain an object in an image, and the object is trained and classified via deep learning. CNN and GAN networks are used to classify the objects to be identified, a determining axis and an identification area are set dynamically according to a classification result, and vehicles are identified and counted. A higher counting precisionis realized in different environments, the method is adapted to model training under the condition that training samples are not rich, data features can be extracted more accurately, and the model classification accuracy is improved.

Description

technical field [0001] The present invention designs a traffic statistics method, specifically a traffic statistics method that combines deep learning and background difference method. Background technique [0002] With the rapid development of social economy, the demand for transportation is increasing day by day, urban traffic congestion and accidents occur frequently, and the traffic environment is deteriorating day by day. Both developed and developing countries are plagued by worsening traffic problems. The traditional way to solve traffic problems is to build or expand roads to increase the carrying capacity of the road network. However, as the population grows, there is less and less space for road construction, and the speed of expanding the road network is far slower than that for traffic. The growth rate of demand. At the same time, the transportation system is a complex and comprehensive system, and it is difficult to solve traffic problems simply from the perspe...

AI Technical Summary

Problems solved by technology

The method of infrared detection can provide a large amount of traffic management information, but its anti-noise ability is not strong, and sometimes it may be necessary to reduce reliability to improve sensitivity; the technology of ground induction coil detection is relatively mature, with stable performance and very accurate counting, it can detect Traffic flow, road occupancy rate, etc., but this method needs to lay coils on the road line, which affects the life of the road surface, and is easily damaged by heavy vehicles; the ultrasonic detection method determines the passing of vehicles according to the time difference between received and returned ultrasonic waves, and has a volume Small, easy to install and other advantages, but it is greatly affected by weather and temperature; the acoustic detection method detects the vehicle by detecting the sound inside the vehicle and the sound of the vehicle contacting the ground, but this method is difficult to detect the stopped vehicle, and sometimes it will miss check vehicle

Images