Kuzushiji Dataset: Historical Japanese Document Pages with Character Annotations

Page images from historical Japanese documents (Kotenseki) are provided with character-level bounding box annotations for Kuzushiji (cursive Japanese) recognition. The dataset was uploaded by Kotomiya07 and was last updated on March 10, 2026.

Use Cases

• Train object detection models for character localization based on bounding box annotations.
• Develop OCR systems for historical Japanese cursive script (Kuzushiji).
• Analyze page layout and character distribution in historical Japanese documents.

Strengths

• Contains character-level bounding box annotations, enabling precise localization tasks.
• Focuses on historical Japanese cursive script (Kuzushiji), a specialized domain.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

huggingface

Freshness

Last updated 2026-03-10 15:45:41; freshness should be verified.

Geography

Japan