Classification of picture art style based on VGGNET

Abstract

Abstract Aiming at the classification of various painting art styles in modern society, a method based on convolution neural network is proposed. The classification of various painting art styles is completed by using convolutional neural network. In VGG-19 (visual geometry) Based on Group-19 (Group-19) network, this paper proposes a picture style recognition algorithm based on visual geometry group 19 mixed transfer learning model. By using the feature extraction ability of convolution neural network, the convolution layer parameters of VGG-19 model on the source data are transferred to the target data model, and then the sparse automatic encoder SAE (sparse auto) is used Finally, softmax classifier is used to classify the artistic style of pictures. The artistic style of painting can be roughly divided into: characters, flowers and birds, landscape. The VGG-19 model is used to predict the three kinds of picture art styles, and then the proportion of the predicted results is compared with the actual naked eye proportion, and the conclusion that VGG-19 model is more suitable for classification of painting art styles is obtained.