Non-motor vehicle re-identification method and device, and computer storage medium

Abstract

The invention provides a non-motor vehicle re-identification method and device and a computer storage medium. The method comprises the steps of determining a non-motor vehicle picture and a driver picture in a to-be-detected picture; determining a first feature vector of the non-motor vehicle picture and a second feature vector of the driver picture; fusing the first feature vector and the secondfeature vector to obtain a feature vector of the to-be-detected picture; and comparing the feature vector with a picture feature vector in a database to obtain a re-identification result. Therefore, according to the embodiment of the invention, when a non-motor vehicle is re-identified, the feature vector of the driver on the non-motor vehicle is considered at the same time, so that the obtained feature vector of the picture where the non-motor vehicle is located is more accurate, and the obtained re-identification result is more reliable.

Description

technical field [0001] The present invention relates to the field of image processing, in particular to a method, device and computer storage medium for re-identification of non-motor vehicles. Background technique [0002] In the application of video structuring, re-identification (re-identification, ReID) of objects with the same identification (ID) is very important. Re-identification (ReID), also known as re-identification, can be used in security, criminal investigation and other related fields. It is mainly used to find one or several images that are most similar to the target in a set of images. The targets can be various objects, such as pedestrians, cars, etc. [0003] When the target is a non-motor vehicle, currently when non-motor vehicle re-identification is performed, feature extraction will be directly performed on the detection area of ​​the non-motor vehicle, which generally includes part of the driver's body parts, because the driver's clothes , posture, e...

AI Technical Summary

Problems solved by technology

[0003] When the target is a non-motor vehicle, currently when non-motor vehicle re-identification is performed, feature extraction will be directly performed on the detection area of ​​the non-motor vehicle, which generally includes part of the driver's body parts, because the driver's clothes , posture, etc. will cause great interference to the detection area of ​​non-motor vehicles, and then cause the accuracy of non-motor vehicle re-identification to be low

Images