Oracle Bone Inscriptions Multi-modal Dataset (OBIMD)
收藏资源简介:
甲骨文多模态数据集(OBIMD)由安阳师范学院甲骨文信息处理教育部重点实验室等机构创建,包含10,077件甲骨文的详细标注信息。该数据集涵盖像素级对齐的拓片和摹本,标注了检测框、字符类别、转录、对应的铭文组及阅读顺序,适用于甲骨文字符检测与识别、拓片去噪、字符匹配等多项AI研究任务。数据集的创建过程结合了自动化的字符注册算法和人工验证,确保了标注的准确性和全面性。该数据集旨在推动AI技术在甲骨文研究领域的应用,解决甲骨文解读中的难题。
The Oracle Bone Multimodal Dataset (OBIMD) was created by the Key Laboratory of Oracle Bone Information Processing at Anyang Normal University and other institutions, containing detailed annotations for 10,077 oracle bones. The dataset encompasses pixel-registered rubbings and copies, annotated with detection boxes, character categories, transcriptions, corresponding inscriptions groups, and reading order, and is suitable for a variety of AI research tasks such as oracle bone character detection and recognition, rubbing denoising, and character matching. The creation process of the dataset integrated automated character registration algorithms with manual verification to ensure the accuracy and comprehensiveness of the annotations. This dataset aims to promote the application of AI technology in the field of oracle bone research and address the challenges in the interpretation of oracle bones.

- 1Oracle Bone Inscriptions Multi-modal Dataset安阳师范学院甲骨文信息处理教育部重点实验室, 腾讯优图实验室, 厦门大学多媒体可信感知与高效计算教育部重点实验室, 腾讯可持续社会价值事业部数字文化实验室, 剑桥大学 · 2024年



