Auto Image Classification Based on Convolution Neural Network

Aiming at the low accuracy of vehicle image retrieval algorithm based on deep learning, an improved vehicle image classification retrieval model based on convolutional neural network is proposed. According to the complexity of the car image, using convolution neural network to extract the image features from Stanford Cars Dataset database, and use a local feature aggregation descriptor (vector of locally aggregated descriptors, VLAD) to represent a picture. Finally, SVM is used to classify the image of the car. The experimental results show that compared with the traditional visual feature classification algorithm, the accuracy of the model is higher and the retrieval effect is better.