Guide updated:
Extract editable text from an image and review recognition mistakes before using it.
Open this guide using the “How to use this tool” link below the tool name. It includes steps, screenshots and videos for computers and phones.
Before you start
Prepare a sharp image with readable text. Straight lines, even lighting and a suitable recognition language improve the starting conditions.
Try with a demonstration file — text-card.svg
These small files were created for this guide and contain no personal data.
Step-by-step instructions
- Choose the picture containing the text.
- Select the recognition language or other available OCR options.
- Start recognition and wait for the text result. Review a displayed charge or confirmation if requested.
- Compare names, dates, numbers and punctuation against the image and correct mistakes.
- Copy the text or use the offered export, keeping the original image for reference.
On a computer
Use the full-width working panel to compare the source, settings and result. Keep the original file so you can try another setting without replacing it.
On a phone
Choose the source from Files or your device file manager. Scroll down to the settings and result; keep this tab open until the operation finishes.
In Safari on iPhone, open the sharing menu and choose Save to Files if the result opens in a new tab. In Android Chrome, check Downloads. Keep this tab open until processing finishes.
For a better result
Crop away unrelated areas before recognition when practical. A larger image cannot compensate for severely blurred letters.
Screenshots and video
Desktop demonstration
The recording shows the English interface. Follow the translated steps above; the controls are in the same order. This recording shows the source and available settings before the main action. Follow the steps above to start processing and save the result.
Phone demonstration
The recording shows the English interface. Follow the translated steps above; the controls are in the same order. This recording shows the source and available settings before the main action. Follow the steps above to start processing and save the result.
Questions and troubleshooting
Why are numbers or names wrong?
Similar characters and uncommon words can confuse recognition. Check important values manually against the image.
Does OCR preserve the page layout?
Plain-text output may not preserve columns, tables or fonts. Review the actual output options rather than assuming a document copy.
If you see a message that the service is overloaded, wait and try again later. A failed attempt does not mean that your source file has changed.
