An Enhanced Neural Network Model Based on VGG-16 for Accurate Recognition of Oracle

Oracle bone inscriptions are ancient Chinese characters. The study of oracle bone inscriptions heavily relies on character identification. At present, the identification of oracle bone inscriptions on rubbings relies on manual annotation by experts. This is a complex and demanding task, requiring a great deal of labor and time. Therefore, automatic image recognition of oracle bone inscriptions is of great value. By fusing the traits of the VGG-16 model and the Inception structure, this paper suggests an enhanced multi-scale convolutional neural network model. Through the experiment on OBC306 data set, compared with the traditional VGG-16 and GoogleNet models, the accuracy of the suggested model in the training set and test set is better than other models, which demonstrates the correctness and superiority of the suggested model.