Convolution neural network for unbalanced vehicle type recognition

The convolution neural network in deep learning has been widely used in image recognition in recent years.For the traditionalvehicle type recognitionmethod by the vehicle scene and the bottom of the feature extraction and other limitations, has been unable toapractical application of the recognitionof the problem.Using the GoogLeNet network of the deep learning Caffe framework and the supervised Kohonen network with the traditional method, the vehicle image data sets (unbalanced vehicle type original image) and vehicle size feature data sets for network training and testing.The accuracy of the two models is compared and verified, and the applicability of the two methods is compared and analyzed.The experimental results show that the classification method of convolution neural network hasahigh accuracy of vehicle type recognitionunder the condition of large data volume.