Documents, retail scenes, signage and handwriting.
Dense mixed-script text (kanji, kana, Latin), vertical layout and unique signage break models trained on Western imagery.
Describe the data you need. We will respond with feasibility, rights approach and next steps.
Scanned and photographed business documents, forms and receipts with mixed kanji, kana and Latin text and vertical layout.
Handwritten samples across ages and styles, including cursive and forms-style writing, from consenting writers.
Shelf and product-display imagery from Japanese retail environments, captured with store permission.
Street, station and storefront signage in natural scenes for scene-text detection and multimodal grounding.
Dish, ingredient and menu imagery with culturally accurate naming, which generic English captions often miss.
Licensed illustration and manga page data via publisher and creator agreements; no data is included without explicit rights.
Bridge, tunnel and building inspection imagery from operators with rights to share.
Not here? Describe the data you need. We source beyond this list.