Kuzushiji Character Dataset Ogihan V1: Historical Japanese Character Images

Kuzushiji Character Dataset (Ogihan / Ogi Domain) contains single-character Kuzushiji image crops derived from Ogihan historical materials. It is published in a format compatible with datasets from the Center for Open Data in the Humanities (CODH). The dataset was created by DimV-Ai and last updated on January 11, 2026.

Use Cases

• Train Kuzushiji OCR models based on single-character image crops.
• Perform character-level recognition tasks on historical Japanese documents.
• Train multimodal and vision-language models using historical character images.
• Conduct comparative research with existing CODH datasets.

Strengths

• Derived from Ogihan historical materials, providing a specific source context.
• Published in a format compatible with existing CODH datasets, facilitating comparative work.
• Last updated on January 11, 2026, suggesting recent maintenance.

Limitations

• Dataset size, row count, and file formats are unknown, limiting suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Ogihan (Ogi Domain) historical materials.

Collection Method

Derived single-character image crops from source materials.

Freshness

Last updated 2026-01-11 13:31:25.

Geography

Japan (Ogihan/Ogi Domain).