Vehicle re-identification method and system based on multidirectional information and multi-branch neural network

Abstract

The invention provides a vehicle re-identification method and system based on vehicle direction information and a multi-branch neural network, and the method comprises the steps: collecting a plurality of to-be-identified pictures of a vehicle, and retrieving a plurality of vehicle comparison pictures on a data set; obtaining direction information of the vehicle to-be-identified pictures and the vehicle comparison pictures; pairing the vehicle to-be-identified pictures with the vehicle comparison pictures to form a plurality of picture groups, and dividing the plurality of picture groups intoa shared vision field group and a non-shared vision field group according to the direction information; inputting the pictures into a training model to obtain shared vision field group features or non-shared vision field group features; calculating the Euclidean distance between the vehicle to-be-identified picture and the vehicle comparison picture according to the shared vision field group features or the non-shared vision field group features, sorting the Euclidean distance, and retrieving a plurality of vehicles having the top similarity with the to-be-identified vehicle; learning two different features according to whether the shared vision field exists or not, so that the features with high discriminability can be learned, the retrieval sorting performance is enhanced, and the accuracy of vehicle re-identification is improved.

Description

technical field [0001] The invention relates to a vehicle re-identification method and system based on multi-directional information and a multi-branch neural network, belonging to the technical fields of computing vision and artificial intelligence. Background technique [0002] With the development of society and economy, the number of vehicles is increasing day by day, and the management of vehicles is also becoming more and more difficult. Vehicle re-identification refers to the process of matching vehicle pictures under different surveillance cameras without relying on license plate information, and finding the target vehicle in the videos captured by non-overlapping cameras at different times. Vehicle re-identification has important applications in real life, such as criminal investigation, urban computing, public management, and intelligent transportation. [0003] Initially, vehicle identification was mainly carried out through some sensors, such as: magnetic sensor...

AI Technical Summary

Problems solved by technology

In addition, although there are some methods that utilize these local details, they ignore the impact of vehicle orientation on feature extraction.

Due to the different shooting angles of the camera and the different driving conditions of the vehicle, the direction of the vehicle on each picture taken is very different.

Images