Awesome
A macOS Menu Bar Application for Text and Math Formula Recognition Based on Pix2Text
简体中文
Note: This application is for macOS only
Updates
- 2024-04-26 Added Auto Recognition (supports three modes) feature to solve the issue of manually clicking recognition button in long-term, multiple recognition scenarios
- Added Auto Recognition (math formula recognition only) feature
- Added text-only recognition function and button
- Application icon no longer shows in dock after launch
- Fixed program freezing issue during long recognition processes
- Updated OCR recognition model to enhance handwritten math formula recognition
- Added mixed recognition for math formulas and text documents
Screenshots
- Test Results
- Menu Bar Application Interface
Installation
git clone https://github.com/horennel/LaTex-OCR_for_macOS.git
pip3 install -r requirements.txt
python3 setup.py py2app -A
- Move the generated
MyLatexOCR.app from the dist folder to your Applications folder
Usage
- Launch
- Start
MyLatexOCR, you'll see the application icon in the menu bar
- Screenshot
- Use any screenshot tool (e.g.,
Snipaste) to capture and copy to clipboard
- Recognition
- Math Formula Only
- Click
Formula OCR button
- You'll receive a notification upon successful recognition
- Mixed Math Formula and Text
- Click
Mixed OCR button
- You'll receive a notification upon successful recognition
- Text Only
- Click
Text OCR button
- You'll receive a notification upon successful recognition
- Notifications can be disabled in system settings
- After receiving notification, you can paste the LaTeX formula and text anywhere
- <span id="auto-ocr">Auto Recognition (supports
Math Formula Mode, Mixed Mode, Text-Only Mode)
- Enable/Disable Auto Recognition
- Click
Auto On/Off button to toggle auto recognition
- Dark menu bar icon indicates auto recognition is enabled
- Light menu bar icon indicates auto recognition is disabled
- Usage
- After enabling auto recognition (defaults to
Math Formula Mode), click the corresponding mode button
- Selected mode button becomes unclickable with a
√ prefix
- No further clicks needed, program automatically recognizes clipboard images
- Important Notes ⚠️
- Program automatically processes clipboard images but cannot determine if recognition is needed
- Copying unneeded images with auto recognition enabled wastes system resources
- Recommended to enable auto recognition only when continuous recognition is needed
Important Notes
- First launch will download models and configuration files, causing longer initial startup time
- Model and config files are stored in
～/.cnstd, ~/.cnocr, and ~/.pix2text
- Application depends on Python environment used during packaging. Repackaging needed if environment changes
Credits for Icon
Development Resources
Thanks to Open Source Projects