Improvement of Oracle Bone Inscription Recognition Accuracy: A Deep Learning Perspective

Xuanming Fu,Zhengfeng Yang,Yidan Zhang,Qianting Zhou,Zhenbing Zeng

doi:10.3390/ijgi11010045

Abstract

Deep learning techniques have been successfully applied in handwriting recognition. Oracle bone inscriptions (OBI) are the earliest hieroglyphs in China and valuable resources for studying the etymology of Chinese characters. OBI are of important historical and cultural value in China; thus, textual research surrounding the characters of OBI is a huge challenge for archaeologists. In this work, we built a dataset named OBI-100, which contains 100 classes of oracle bone inscriptions collected from two OBI dictionaries. The dataset includes more than 128,000 character samples related to the natural environment, humans, animals, plants, etc. In addition, we propose improved models based on three typical deep convolutional network structures to recognize the OBI-100 dataset. By modifying the parameters, adjusting the network structures, and adopting optimization strategies, we demonstrate experimentally that these models perform fairly well in OBI recognition. For the 100-category OBI classification task, the optimal model achieves an accuracy of 99.5%, which shows competitive performance compared with other state-of-the-art approaches. We hope that this work can provide a valuable tool for character recognition of OBI.

Highlights

Oracle bone inscriptions (OBI) were recorded from as early as the Shang Dynasty inChina [1]
We visualized the changes in the training loss value, the accuracy rates on the training set and the validation set of different models during the training process as the number of training epochs increases
OBI-100 can fill the gap of publicly available datasets in the applications of deep learning in OBI research

Summary

Introduction

Oracle bone inscriptions (OBI) were recorded from as early as the Shang Dynasty inChina [1]. Convolutional neural network (CNN) [21] is a multi-layer feed-forward neural network that can extract features and properties from the input data. As shown, the basic CNN structure consists of an input layer, several convolutional layers, and pooling layers, as well as several fully connected layers and an output layer. The convolutional layer is designed to extract features from the input data, which contain many convolutional kernels. The parameters of the convolutional layer include the size of the kernel, the step size, and the padding method [23] These three factors jointly determine the size of the output feature map of the convolutional layer [24]. After feature extraction [26] by the convolutional layer, the output feature map is transferred into the pooling layer for feature selection and information filtering. The output layer uses a logistic function or a normalized exponential function to output the classification label or probability

Methods

Results

Conclusion