OnScreenOCR

A powerful graphical on-screen optical character recognition (OCR) tool using Tesseract. I made this tool because I found places where I would like to improve existing tools like Microsoft PowerToys' "Text Extract" feature.

Not my best logo design... I needed something for the system tray, though!

1. Features
2. Pictures
3. Installation
4. Configuration
5. TODO
6. Development
7. Creating the installer

Features

The features of OnScreenOCR include:

• Fully GPU-accelerated rendering using wgpu
• Multi-platform support: Windows, MacOS, and Linux (untested at the moment, some features TODO on other platforms)
• Live preview of the OCR result
• Support for taking screenshots
• Support for multiple OCR languages
• Result fixing and reformatting
  • Reformat to remove hyphens from end of lines, moving the word to fit entirely on the line
  • More to come in the future! If you have any suggestions or find common formatting failure points/annoyances, please open an issue!
• Ability to copy without newlines
• Ability to fine-tune Tesseract's parameters
  • Ability to export in other Tesseract formats (TSV, Alto, HOCR)
• Support for non-rectangular selections
• Support for multiple monitors
• Keybinds for common actions (Ctrl+C to copy, Ctrl+Z to undo, arrows to move selection, etc.)
• Full undo/redo history
• Stays in system tray when closed
• Numerous intuitive selection-related interactions, including drawing outlines, shifting edges/vertices, removing edges/vertices, and more.
• An installer that allows you to automatically start the application on boot

Pictures

Installation

You may download the application from the releases page.
Currently, I only distribute x64 Windows binaries (since support for other platforms is untested and some features are not implemented on other platforms).
I distribute an installer (OnScreenOCR.exe) and a standalone executable (OnScreenOCRStandalone.exe).
The application will create an icon in your tray while running. The default keybind is Shift + Alt + Z, but you may change this within the overlay.

Configuration

Configuration files are stored under your user's configuration directory.
On Windows, this is {FOLDERID_RoamingAppData}\OnScreenOCR (usually C:\Users\<username>\AppData\Roaming\OnScreenOCR). On MacOS, this is ~/Library/Application Support/OnScreenOCR.
On Linux, this is $XDG_CONFIG_HOME/OnScreenOCR or $HOME/.config/OnScreenOCR.

Configuration files include:

• settings.bin: A binary file containing the application's simple settings, configurable from the application. (e.g. preserve newlines or background blur enabled)
• tesseract_settings.toml: A TOML file containing the Tesseract settings. When changing this file, the "reload" button must be clicked in the application. (e.g. OCR language, OCR parameters)
• tessdata: A directory containing the Tesseract data files. This directory is created when the application is run for the first time with a few default languages. To add more languages, simply copy the .traineddata files into this directory and add the language to the tesseract_settings.toml file. Configuration is documented in the file.
• correction_data: A directory containing the correcton data files. There is one subdirectory per required dictionary for correction. If no file exists for a language, the correction will not be applied.
  • Currently, there's only one type of correction -- the "hyphenated" correction. For every language you add that requires end-of-line hyphen correction, you should add a dictionary of hyphenated words to correction_data/hyphenated/{language code}.txt, with one word per line.

TODO

• Add support for MacOS and Linux
• Better documentation on how the interaction system, configuration, and probably other parts of the application work

Development

Since the OCR dependency used (Leptess) relies on vcpkg dependencies, you need to run the following (and clone with submodules!):

# To install the LLVM (clang)
winget install LLVM

# To install vcpkg dependencies
.\vcpkg\bootstrap-vcpkg.bat
.\vcpkg\vcpkg integrate install
.\vcpkg\vcpkg install tesseract:x64-windows-static-md
.\vcpkg\vcpkg install leptonica:x64-windows-static-md

NOTE: When installing, vcpkg can't be in a directory with spaces in the path!

On platforms other than Windows, follow the instructons in Leptess' README here.

Finally, you can run the project:

cargo run

Creating the installer

0. Install Inno Setup 6
1. Build the program with cargo build --release and move the executable from target/release/OnScreenOCR.exe into this folder. Make sure it is called OnScreenOCR.exe.
2. Open OnScreenOCR.iss in Inno Setup Compiler
3. Change MyAppVersion to the current app version.
4. Run Build -> Compile
5. The installer will be placed in a folder called Output

installer/pre-install_message.txt contains the message that is displayed before installation.