Vehicle identification method and device

Abstract

This application provides a vehicle identification method and device. The method includes the following steps that: a vehicle image to be identified is obtained; a first deep learning network which is obtained through pre-training is utilized to identify the vehicle image to be identified; the network structure of the first deep learning network includes convolutional layers, pooling layers and all-connection layers, the pooling layers are connected behind the convolutional layers, the all-connection layers are connected behind the pooling layers, and each output node on the all-connection layers is the vehicle attribute probability of the vehicle image; and the vehicle attribute information of the vehicle image to be identified is determined according to the vehicle attribute probability. According to the method provided by the technical schemes of the invention, since the deep learning network is utilized to identify a vehicle, the deep learning network is competent enough to describe and distinguish objects, and therefore, compared with a method according to which features are defined manually to carry out classification, the method of the invention has higher accuracy as well as lower false positive rate and false negative rate.

Description

technical field [0001] The present application relates to the technical field of computer vision, and in particular to a vehicle recognition method and device. Background technique [0002] At present, when identifying specific content in a picture, the following steps are usually included: [0003] The first step is to detect the position of the object of interest in the picture. For example, if you want to recognize the vehicle, you need to use a detector to find the car from the picture. The output of the detector is that the car is on the picture. coordinate of; [0004] The second step is to cut the car from the original picture according to the coordinate position, and put the cut picture into the classifier, and the output result of the classifier is the recognition result of the car. [0005] In the second step, the input original image pixel values ​​are usually converted into artificially defined features (human-engineered features), such as: scale-invariant feat...

AI Technical Summary

Problems solved by technology

[0008] The embodiment of the present application proposes a vehicle recognition method and device to solve the technical problem of low accuracy in object recognition by the object recognition method in the prior art

Images