A vehicle fine-grained classification method and device

Abstract

The present invention provides a vehicle fine-grained classification method and device. The classification method includes: extracting the basic feature map of the input picture through a convolutional neural network; adaptively constructing a global structure map according to the basic feature map; using the graph convolutional neural network to The structural map performs association relationship reasoning to form global guidance information; the global guidance information is applied to the basic feature map to obtain an enhanced feature map; the enhanced feature map is input into the classifier for classification. The classification method of the present invention performs association relationship reasoning between different parts of the vehicle, and applies the relationship reasoning results to the basic feature map, so that the obtained feature map has stronger expressive ability and improves the accuracy of vehicle fine-grained image classification. In addition, this classification method does not need to label the key areas, and the feature extraction is fast, the application scenarios are wide, and the calculation consumption is low.

Description

technical field [0001] The invention relates to the technical field of vehicle classification, in particular to a fine-grained vehicle classification method and device. Background technique [0002] As an important means of transportation for human travel, vehicles are closely related to people's lives, and the development of modern vehicle production technology has also made more and more types of vehicles appear in people's daily life; vehicles usually have similar appearance structures, Different types of vehicles only have slight differences in the positions of lights, doors, etc., and fine-grained identification of vehicles often requires experts in related fields. [0003] The fine-grained classification of vehicles is one of the most important tasks in computer vision. It not only plays an important role in other vision tasks, such as vehicle re-identification, vehicle tracking, but also provides assistance for the city's brain and intelligent traffic construction; at...

AI Technical Summary

Problems solved by technology

[0013] In view of this, the purpose of the present invention is to propose a vehicle fine-grained classification method to solve the problems existing in the existing classification methods that the key area information needs to be marked, the key areas are not related, and the calculation consumption is large

Images