Open-sourcing of the World's Largest Oracle Bone Multimodal Data Set
TapTechNews July 5th news, the Digital Oracle Bone Co-creation Center officially open-sourced the world's largest multimodal oracle bone data set today, which contains a total of 10,000 oracle bone rubbings, facsimiles, the corresponding positions, corresponding headers, corresponding interpretations, and data such as the grouping of inscriptions and the order of interpretations.
According to the introduction, all researchers can develop algorithms such as oracle bone detection, recognition, facsimile generation, glyph matching, and interpretation based on this data set to accelerate the intelligent process of oracle bone research.
TapTechNews learned that the Digital Oracle Bone Co-creation Center was jointly initiated by Anyang Normal University Oracle Bone Information Processing Ministry of Education Laboratory, Tencent SSV Digital Culture Laboratory, Tencent YouTu Laboratory, Chinese Academy of Social Sciences Oracle Bone Studies and Yin-Shang History Research Center, Chinese Academy of Social Sciences Institute of Archeology Anyang Workstation, Xiamen University Multimedia Trusted Sensing and Efficient Computing Ministry of Education Key Laboratory, Zhengzhou University Chinese Character Civilization Research Center and other units, and has received the support of global universities and research institutions such as the Institute of Ancient History of the Chinese Academy of Social Sciences, the University of Cambridge in the UK, the École pratique des hautes études in France, Ritsumeikan University in Japan, Rutgers University in the US, and the University of California, Los Angeles.
Tencent YouTu Laboratory, Tencent SSV Digital Culture Laboratory, Xiamen University, and Anyang Normal University jointly developed AI model technologies:
Oracle bone character detection model: The annotation accuracy rate is over 90%.
Facsimile generation model: Pixel-wise alignment between facsimile and rubbing.
Glyph matching model: Automatically match similar characters.
Oracle bone duplicate checking model: Realize facsimile deduplication and rubbing tracing in a large number of rubbings and facsimiles.
The world's largest multimodal oracle bone data set has been launched on the Oracle Bone AI Collaboration Platform, and this platform can also query oracle bones and oracle bone pieces information. Specific functions can be accessed and experienced by yourself:
https://www.jgwlbq.org.cn/home
