Umi-OCR
Free, open-source OCR app that reads text from screenshots, images, and PDFs, offline.
- Stars
- 48.4k
- Forks
- 4.7k
- Licence
- MIT
- Language
- Python
- Last update
- 12 months ago
- Type
- App
What it is
Umi-OCR is a desktop program that turns text in images into editable text. You can capture part of your screen, drop in many image files at once, or load PDFs and scanned documents. It runs offline, so your files are not uploaded. The README also describes QR and barcode reading and generation.
- Capture a region of the screen with a shortcut and copy the recognized text
- Batch-import hundreds of images and save results as txt, jsonl, md, or csv
- Turn a scanned PDF into a searchable PDF or extract its text
- Read QR codes or barcodes from images, or generate QR code images from text
Great for
- People who need to copy text out of screenshots or scanned pages
- Users who want to batch-process many image files without a network connection
- Anyone working with PDFs that are scans and need a searchable copy
- Users who want to remove watermarks or headers and footers from OCR output using ignore regions
Not ideal for
- Users on platforms other than Windows and Linux, since the README lists only those (Windows 7 x64 and Linux x64 are mentioned), because other systems are not covered
- People who need cloud-based or handwriting-focused recognition, since the README does not describe those features
- Users who want a pure command-line-only tool, since the GUI is the main interface described
How to get it
- 1
Download the app
Clone the repository
git clone https://github.com/hiroi-sora/Umi-OCR - 2
Run it
- Download the release package from the GitHub releases page or the mirror links in the README
- Extract the .7z or .7z.exe archive
- Run Umi-OCR.exe on Windows (the README says no installation is needed)
- Open a tab such as 截图OCR (screenshot OCR) and use the shortcut to capture text
- 3
Get the most out of it
- Use 文本后处理 layout options such as 多栏-按自然段换行 for multi-column pages
- Set ignore regions in batch OCR to exclude watermarks
- Adjust 限制图像边长 in page settings for very large images, per the README
- Use the command-line or HTTP interfaces documented in docs/README_CLI.md and docs/http/README.md for automation
Downloads
Strengths
- Free and open source under the MIT license, with the README stating the code is fully free
- Works offline; the README says no network is needed after download
- Supports batch processing and multiple export formats
Trade-offs
- The README is mostly in Chinese and the main docs are partly in other files, so setup details may be harder to follow for some readers
- Some features depend on the chosen OCR engine (Rapid or Paddle), and the README does not compare their accuracy
- Vertical text support depends on the OCR engine itself, per the README
How it compares
| Project | Stars | Type | Level | Updated | Best for |
|---|---|---|---|---|---|
| Umi-OCR (this) | 48.4k | App | beginner | 12 months ago | People who need to copy text out of screenshots or scanned pages |
| zerox | 12.3k | Library | intermediate | 1.5 years ago | Developers building pipelines that feed documents into AI or search systems |
| Captura | 10.9k | App | beginner | 6 months ago | Windows users who want a free screen recorder with no ads or paid tier mentioned in the README |
Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.