Dense vehicle detection method based on vehicle counting perception attention

Abstract

The invention discloses a dense vehicle detection method based on vehicle counting perception attention, which is characterized in that vehicles in a dense environment are detected based on a deep learning network model, the network model comprises a vehicle number perception network model and a dense target detection network model, and the two network models are integrated through a vehicle counting perception attention module. Effective features of vehicle detection are enhanced, semantic information and spatial information of vehicles are effectively represented by constructing a vehicle counting density map, a vehicle number perception loss function based on cross entropy is calculated, the model is supervised to learn the semantic information and the spatial information of the vehicles, and the vehicle detection effect of the network model is improved.

Description

technical field [0001] The invention relates to the technical field of vehicle detection, in particular to a dense vehicle detection method based on vehicle counting perception attention Background technique [0002] With the acceleration of my country's urbanization and automobile popularization, motor vehicles have become the mainstream mode of transportation. The construction speed of urban road infrastructure cannot meet the demand of rapid growth of vehicle ownership, and the resulting problems of traffic safety, road congestion, and environmental pollution have become increasingly prominent. The densely distributed vehicles in the actual scene will not only directly lead to the occurrence of traffic accidents, but also affect the service capabilities of urban infrastructure. How to efficiently analyze densely distributed vehicles to ensure travel safety and smooth roads has become an urgent problem for intelligent traffic analysis systems, which first involves dense v...

AI Technical Summary

Problems solved by technology

On the one hand, these objective factors cause smaller vehicles to lose more appearance details, and it is difficult for the detector to capture the characteristics of all vehicles; on the other hand, it will cause vehicles to overlap together, and it is difficult for the detector to distinguish the position of a single vehicle

[0004] Deep learning technology has strong data fitting ability relying on deep structure, and can comprehensively consider the distribution of visual features of the target. Although the existing vehicle detection technology based on deep learning has made a major breakthrough, most of these technologies are aimed at detecting natural Vehicles in the scene, without considering the challenges brought by complex scenes, it is difficult to directly solve the detection problem of dense vehicles

Existing methods often have problems such as missing detection and multiple detection.

Images