Image to Text
Extract editable text from photos, screenshots and scanned images in your browser; your file is not uploaded.
Your files stay in this browser. No uploads, accounts or saved document data.
Up to 50 queue items, including up to 50 PDF pages. Images above 40 megapixels are reduced before OCR. GIF: first frame only.
Printed text works best. Handwriting, blurred or very small text, tables and complex layouts may be read poorly or in a different order. Language files download on first use (one to a few MB each); your browser may cache them.
Most selected pages in a PDF already contain text. You can continue with OCR or extract the existing text directly: Open PDF to text
Review and save
Confidence is the engine's own estimate, not a measure of correctness. Compare important details with the original.
Copy all and Download TXT use this edited text. Editing an item, adding or removing items, or recognising again rebuilds this field from the item texts.
Manual text corrections are included in TXT only. The searchable PDF keeps the engine's original text layer; compare it with the page images.
How to recognise text in an image
- To turn an image to text, choose a file, drop it onto the page, paste a screenshot or use Take a photo.
- Select the document language and, if needed, a second language. The tool does not detect languages automatically.
- Rotate the preview or drag over the part you want to read. You can return to the whole image.
- Choose a layout for paragraphs, one block, one line or sparse text, then start recognition.
- Review and edit the result beside its preview. Copy it, download TXT or request a searchable PDF before recognition.
Useful ways to extract text
- Turn a clear photo to text when you need to reuse a printed notice or receipt.
- Convert a screenshot to text so you can copy a message or label.
- Crop one column of a scanned page to help the engine follow its reading order.
- Process several images in one queue and copy their combined text afterward.
- Try recognition again after changing the language, rotation, crop or image improvement setting.
Image OCR questions
How reliable is the recognised text?
Clear, well-lit printed text usually works best. Blur, tiny letters, handwriting and complex tables can cause errors. The displayed confidence is the engine's estimate; check the editable result against the image.
Which languages can I choose?
There are 20 document languages. Choose one required language and optionally a second; selecting the languages actually present can help recognition. The tool does not identify them for you.
Does my image leave my device?
Recognition runs locally in your browser, and the file is not uploaded. Language files download on first use and may be cached. The site displays ads.
How many files can I process?
A run accepts up to 50 queue items. Images above 40 megapixels are downscaled before recognition, and large files also depend on available device memory.
What can I save?
You can copy text or download a TXT file. Optional searchable PDF output keeps the page images alongside searchable text, so its file size can be large.
Can I use a scanned PDF here?
Yes. PDF pages can be recognised with the same tool. For page selection and guidance about an existing text layer, use the PDF OCR page.