ToolRobinNo account needed

Image / PDF Text Extractor

Extract and review text from screenshots, photos and PDFs in your browser.

Choose images or one PDF, select the text language and extract. Review each page beside the original, edit any mistakes, then copy or download TXT.

How to use

Extract text from a screenshot or PDF

Extract and review text from screenshots, photos and PDFs in your browser.

Step by step

  1. Provide the source using: Choose images or a PDF.
  2. Set: Text language / Recognition mode. Use the reference below to choose the options that fit your task.
  3. Choose: Extract text. Inspect the result before copying, downloading or using it elsewhere.
  4. Choose images or one PDF, select the text language and extract. Review each page beside the original, edit any mistakes, then copy or download TXT.

Choosing the settings

OptionWhat changes
Text languageUp to 10 images or one PDF with 20 pages. Printed text works best. Tables, handwriting and complex columns need manual review. Bengali recognition is not supported.
Recognition modeThe engine and selected models download when needed. Files stay in this browser. Quality mode loads about 31 MB of models plus the engine; lightweight mode loads about 6 MB of models. Cached assets can be reused.
Recognize this page againExisting PDF text is extracted directly. Scanned pages use OCR. Recognize this page again forces OCR if the original text layer is incomplete. Export is plain text, not a searchable PDF or an Excel table.

Example and expected result

Example: the sample invoice contains “Total: 125.50 USD”. Check the numbers against the original before copying. TXT keeps text and page separators; it does not preserve tables or PDF layout.

If it is not working

  • Check Arabic, Urdu and Vietnamese results carefully: mixed scripts or accents can cause missing numbers or letters.
  • When recognition is wrong, check the selected script and compare the affected line with the image. Correct names, decimal points and digits before copying. Enlarging a blurred source or changing the interface language does not supply missing characters.
  • The first model download needs a working connection even though your file is processed locally. If loading fails, retry the missing download; if memory is insufficient, use a smaller input. A model result still needs comparison with the original.

Recognition limits

Up to 10 images or one PDF with 20 pages. Printed text works best. Tables, handwriting and complex columns need manual review. Bengali recognition is not supported.

Check Arabic, Urdu and Vietnamese results carefully: mixed scripts or accents can cause missing numbers or letters.

First download

The engine and selected models download when needed. Files stay in this browser. Quality mode loads about 31 MB of models plus the engine; lightweight mode loads about 6 MB of models. Cached assets can be reused.

PDF text layers

Existing PDF text is extracted directly. Scanned pages use OCR. Recognize this page again forces OCR if the original text layer is incomplete. Export is plain text, not a searchable PDF or an Excel table.

You might also need