Vehicle re-identification method and device based on cascade network

Abstract

The invention provides a vehicle re-identification method and device based on a cascade network, and the method comprises the steps: carrying out the feature extraction of a target vehicle image and all candidate vehicle images, and obtaining the image features of the target vehicle image and the image features of all candidate vehicle images; based on a vehicle type recognition model, performing feature extraction on the image features of the target vehicle image and the image features of the candidate vehicle images, and obtaining the vehicle type features of the target vehicle and the vehicle type features of the candidate vehicles; based on the image feature of the target vehicle image and the vehicle type feature of the target vehicle, determining the identity feature of the target vehicle, and based on the image feature of each candidate vehicle image and the vehicle type feature of each candidate vehicle, determining the identity feature of each candidate vehicle; and based on the target vehicle identity feature and each candidate vehicle identity feature, determining a target vehicle from each candidate vehicle, and obtaining a vehicle re-identification result. According to the invention, the vehicle re-identification precision can be further improved.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to a cascaded network-based vehicle re-identification method and device. Background technique [0002] Given an image of a probed vehicle, the task of vehicle re-identification is to search a database for images containing the same vehicle captured by multiple cameras. Vehicle re-identification has universal applications in video surveillance, intelligent transportation and urban computing, and can quickly find, locate and track target vehicles in large-scale surveillance videos. [0003] At present, the target vehicle image features and multiple candidate vehicle image features are extracted from the target vehicle image to be recognized and the surveillance video image, and then based on the target vehicle image features and each candidate vehicle image feature, the corresponding candidate vehicles are determined. target vehicle. However, when there are multiple candidat...

AI Technical Summary

Problems solved by technology

[0004] The present invention provides a cascaded network-based vehicle re-identification method and device to solve the defect of low vehicle identification accuracy in the prior art

Images