Aiming at the problem of low recognition rate of vehicle model

a vehicle model fine recognition algorithm based on local features and multiple losses is proposed.Firstly

the car face area with large degree of discrimination is obtained by faster region-based CNN（ Faster RCNN）.Then

in the identification network

the car face feature is extracted by VGG 16

and the Soft Max loss and the center loss are weighted and fused at the output layer

effectively guiding the distance within the network learning class to be small.

the characteristics of the distance between the classes is large.The experimental results show that the recognition accuracy of the algorithm in the VRID—1 dataset of Zhongshan University is 98.80 %

and the accuracy of the vehicle model fine recognition has been significantly improved.