Surya — разбор и оцифровка документов

★ 21.3K

Surya is an open-source document-parsing tool: text recognition, page layout analysis, reading-order detection and table parsing. Reach for it when documents must be digitised well and across many languages: recognition covers 91 languages, complex scripts included. Unlike plain letter extraction, Surya understands a page's structure — where the heading is, where the columns are, in what order to read, where a table's borders lie — so the result keeps the source's structure. It helps digitise archives, multilingual documents, forms and scientific material, and serves as the base for your own document pipeline. A compact model underneath runs on ordinary hardware. It goes one way: a document image into structured text.