Application of knowledge graphs and deep learning in automatic art analysis
To improve the accuracy of predicting artistic attributes,a new classification model called Art-Graph is proposed.The model adopts a multi-modal learning strategy and a multi-task learning approach,and applies specific methods during testing.To construct a knowledge graph,WikiArt and DBpedia are used,and a method based on the knowledge graph is proposed for predicting artistic attributes.Experiment results demonstrate the effectiveness and accuracy of the proposed method.
万方数据
维普
