Oracle Bone Inscriptions Detection in Rubbings Based on Deep Learning

Abstract

The rubbings of oracle bone inscriptions are an important carrier of oracle bone inscriptions, which are often used by ancient Chinese language expert. Oracle bone inscriptions detection in rubbings is a basic step for studying oracle bone inscriptions by using computer technology. Because of the remote times, some of the oracle bones have been worn seriously or broken, the content in the rubbings may be interfered by noise. How to effectively detect the oracle bone inscriptions in rubbings is the focus of oracle bone inscriptions research based on rubbings. In this paper we applied YOLOv4 model to the oracle bone inscriptions detection. The experimental results show that the YOLOv4 model can be effectively used to detect the characters in rubbings, but there are still some deficiencies in the ability to distinguish noise.