Skip to content

README.en.md

Latest commit

 

History

History
67 lines (47 loc) · 2.17 KB

README.en.md

File metadata and controls

67 lines (47 loc) · 2.17 KB
Raw

中文 En

GOT-OCR2 Windows GUI

A Windows desktop application for local optical character recognition (OCR) using the GOT-OCR2 model. This application provides a graphical interface for converting images containing text into editable and formatted text.

Here’s a screenshot featuring the image preview at the top, followed by the OCR-extracted markdown/latex text, and the rendered text displayed below.

Features

  • Load and preview images
  • Automatic image resizing and preprocessing
  • OCR processing with GOT-OCR2 model
  • Formatted text output (Markdown and Latex)
  • HTML preview of formatted output
  • GPU acceleration support

Requirements

  • Windows 10/11 (64-bit)
  • .NET Framework 4.8
  • Visual Studio 2022 (for building from source)
  • GPU with Vulkan driver

Installation and Usage

  1. Download the latest release from GitHub Releases
  2. Launch the got-win.exe application
  3. Click "Load Image" to select an image file
  4. The application will:
    • Display the original image
    • Show the preprocessed version
    • Process the image using OCR
  5. View the OCR results in the text box
  6. Use the "Preview" button to see formatted output
  7. Check "Formatted Output" for Markdown formatting

Building from Source

  1. Clone the repository
  2. Install the Vulkan SDK
  3. Get the following files from release file and put in got-win/ dir: libocr.dll, encoder_single.onnx, got_decoder-q4_k_m.gguf
  4. Open got-win.sln in Visual Studio 2022
  5. Restore NuGet packages
  6. Build the solution (x64 platform)
cd got-win
nuget restore
msbuild got-win.sln /p:Configuration=Release /p:Platform=x64

Troubleshooting

  • Ensure the model files are in the correct location
  • Verify GPU drivers are up to date if using GPU acceleration
  • Check console output for any error messages

Acknowledgments