Umi-OCR
hiroi-sora (Umi-OCR project)
Umi-OCR is a desktop text recognition program that runs entirely locally, built on the PaddleOCR and RapidOCR engines. You can select a screen region and get its text, run a batch over a whole folder of images, and process scanned PDF documents into text or into a PDF with a text layer. It has settings for merging lines into paragraphs and for ignoring areas such as watermarks and page headers, plus a command line and an HTTP API for automation. It suits turning an archive of scans, data sheets and old standards into searchable text. Limitation: the built-in engines are strongest in Chinese, English and Japanese, and Hebrew is not supported.
Ratings & Reviews
| Licence type | Completely Free, Open Source |
|---|---|
| Licence terms | Open source under the MIT licence: free for any use, including commercial, with no registration and no internet connection needed. The interface is available in English; Linux is supported through a separate runtime package and the project’s build guide. |
| Platforms | Linux, Windows |
| Engineering field | Architecture, Civil Engineering, Electrical & Electronics, Mechanical Engineering |
| Tool type | Documents & PDF |
| Vendor site | github.com |
This link goes to the vendor site. EngiFree hosts no installers and distributes no cracked versions.
Last verified: 2026-10-16
