Commercial training data from specialist providers. Listings show what each source advertises; licensing is confirmed with the provider before any agreement.
100 People - Handwriting OCR Data of Japanese and Korean,.
500,000 Images - Natural Scenes & Document Photograph Scenes & Electronic Scenes OCR Data of 21 Countries.
An Arabic OCR image dataset for real-world posters and flyers in the media and retail domains, built with JSON annotations containing bounding boxes and language…
An Arabic OCR image dataset for real-world vowelized text images in education and general domains, built with JSON annotations containing bounding boxes and language…
Million Level Document OCR and Structured Analysis Data, Including textbooks, e-books, teaching aids, patents, theses, etc.
Azimov is egocentric (first-person) data recorded by trained professionals doing their actual jobs on live, operating sites.
An English OCR and document parsing sample dataset for manufacturing documents, built with bounding-box JSON annotations for PoC, Document AI, and model evaluation use…
The provider’s Korea EHR Dataset is a patient-consented, multi-institution collection of real hospital records from South Korea - the only Korean clinical corpus of this…
Pakistan's archives, museums, universities and libraries hold centuries of books and handwritten manuscripts in scripts almost no AI model has seen.
A collection of real-world 3d captures (lidar, photogrammetry, and drone), paired with 3d geometry and, on a curated subset, multi-view imagery with per-frame camera…
An Arabic OCR image dataset for real-world food menu and printed document images, built with JSON annotations containing bounding boxes and language metadata.
An Arabic OCR image dataset for real-world magazine cover images in the media and entertainment domain, built with JSON annotations containing bounding-box coordinates.
An Arabic OCR image dataset for real-world back-side product packaging images in the retail and food domains, built with JSON annotations containing bounding-box…
An Arabic OCR image dataset for real-world front-side product packaging images in the retail and food domains, built with JSON annotations containing bounding-box…
An Arabic OCR image dataset for real-world public and transportation signs and notices, built with JSON annotations containing bounding boxes and language metadata.
An Arabic OCR image dataset for real-world retail storefront signs, built with JSON annotations containing bounding boxes, language metadata, and reading direction…
A red-teaming image dataset designed for AI safety evaluation.
A Japanese handwritten text OCR image dataset built from real-world images with JSON annotations containing bounding boxes, language metadata, and reading direction…
A Korean document OCR image dataset containing tables, LaTeX formulas, text, and graphs, built with JSON annotations containing bounding-box coordinates.
A Korean handwritten text OCR image dataset built from real-world images with JSON annotations containing bounding boxes, language metadata, and reading direction…
A multilingual OCR image dataset for menu images in seven languages, including Korean, English, Chinese, and Japanese, comprising original images, translated-text…
The PPT Visual Rendering Dataset is a large-scale collection of 850K+ unique PowerPoint presentation files in PPT and PPTX format, built as AI training data for…
A Thai curved text OCR image dataset built from real-world images with JSON annotations containing bounding boxes and language metadata.
A Thai handwritten text OCR image dataset built from real-world images with JSON annotations containing bounding boxes, language metadata, and reading direction…