Vehicle image recognition method and related device

Abstract

The invention discloses a vehicle image recognition method and a related device. The method comprises the steps of pre-training based on a vehicle image sample marked with a vehicle identifier, a vehicle model label and a vehicle color label at the same time and a convolutional neural network to obtain a target vehicle recognition model; acquiring a to-be-identified vehicle image including the vehicle area image; and identifying the to-be-identified vehicle image by using the target vehicle identification model to obtain a target vehicle identifier, a target vehicle model and a target vehiclecolor of the to-be-identified vehicle image. Correlation among vehicle type identification, vehicle color identification and vehicle re-identification is considered; learning v. Vehicle features, vehicle model features and vehicle color features are learned at the same time through a convolutional neural network to obtain a target vehicle recognition model; the model can realize vehicle re-identification, vehicle model identification and vehicle color identification at the same time, so that the identification algorithm is simplified, the calculation amount is reduced, the calculation resources are saved, the operation environment requirement is reduced, and the integration and miniaturization of the system are realized.

Description

Technical field [0001] This application relates to the field of image processing technology, and in particular to a method and related devices for vehicle image recognition. Background technique [0002] With the rapid development of science and technology, in various scenarios such as road monitoring and analysis, intelligent parking lots, etc., it is necessary to identify the type and color of vehicles to facilitate statistics of passing vehicles. At the same time, since the vehicle appears in the images of multiple different cameras during driving, the same vehicle needs to be identified in the images of different cameras, that is, the vehicle needs to be re-identified. [0003] In the prior art, based on the three different requirements of vehicle type recognition, vehicle color recognition, and vehicle re-recognition, three different neural networks are generally used to perform vehicle type recognition, vehicle color recognition, and vehicle re-recognition, respectively. [00...

AI Technical Summary

Problems solved by technology

[0004] However, the inventor found through research that the above three different neural networks need three completely independent processing units to complete vehicle type recognition, vehicle color recognition and vehicle re-recognition. It consumes more computing resources and has higher requirements on the operating environment, which is not conducive to system integration and miniaturization

Images