Tool to quickly OCR a part of screen

Download here

Please note, this is a hobby project, I use it successfully with no issues, hope some of you might find it useful.

Greg

Description:

screen2text

Select any region of your screen, extract the text with OCR, and paste it anywhere — the result lands in your clipboard automatically.

Uses EasyOCR running entirely on your machine (no cloud, no internet required after setup).


Requirements

  • Windows 10 or 11
  • Internet connection during setup (to download Python and the OCR model)
  • ~500 MB free disk space (Python + EasyOCR model weights)

No GPU required. A GPU (NVIDIA CUDA) will be used automatically if available and will make OCR noticeably faster.


Installation

Step 1 — Copy the files

Place the screen2text folder anywhere you like, for example:

C:\Users\YourName\screen2text\

The folder must contain: screen2text.py screen2text.bat setup.bat requirements.txt

Step 2 — Run setup.bat

Double-click setup.bat.

  • If Python is not installed, the script will install it automatically via winget (requires Windows 10 version 1809 or later). After Python installs, close the window and run setup.bat again to continue.
  • If Python is already installed, setup continues immediately and installs all required Python packages.

The EasyOCR model weights (~100 MB) are not downloaded during setup — they download automatically the first time you run the tool.

Step 3 — First run

Double-click screen2text.bat.

The first run will print: Loading EasyOCR model (CPU)... and download the model. This takes 1–2 minutes depending on your connection. Every run after that starts immediately.


Usage

  1. Run screen2text.bat (or trigger it via a hotkey — see below).
  2. Your screen dims. Drag a rectangle around the text you want to capture.
  3. Press Esc to cancel without copying.
  4. The extracted text is copied to your clipboard. A small popup confirms what was copied.
  5. Paste anywhere with Ctrl+V.

AutoHotkey hotkey (optional)

To trigger screen2text with a keyboard shortcut (e.g. Ctrl+PrintScreen), add the following to your AutoHotkey script file (.ahk):

ahk ^PrintScreen:: Run, cmd.exe /c "C:\path\to\screen2text\screen2text.bat", C:\path\to\screen2text, Hide return

Replace C:\path\to\screen2text with the actual folder where you placed the files.

Example — if your files are in C:\Tools\screen2text\ahk ^PrintScreen:: Run, cmd.exe /c "C:\Tools\screen2text\screen2text.bat", C:\Tools\screen2text, Hide return

If you do not have AutoHotkey installed, download it from autohotkey.com. After editing the .ahk file, right-click it and choose Run Script (or configure it to start with Windows via Task Scheduler or the Startup folder).


Troubleshooting

Problem Fix
winget not found Update Windows or install Python manually from python.org — tick “Add Python to PATH”
Screen goes black but nothing happens Minimise all windows before triggering — the overlay captures the desktop as it was at launch
No text detected Try selecting a larger area or ensure the text is legible; very small or low-contrast text may be missed
Wrong line breaks in output Select a tighter region that excludes surrounding UI elements
GPU not detected Install CUDA Toolkit and the CUDA-enabled version of PyTorch, then reinstall EasyOCR

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *