For the current development build. The public Chrome Web Store link is pending. This guide explains the extension workflow; it does not run the tool on this website. Check installation status.
Before you start
Use an image or screenshot containing clear, horizontal printed text. OCR models are included in the extension, so recognition can work offline from the first use. Obtaining a source image from a website may still require a network connection and access to that page.
The extension’s Image OCR tool runs on your device. Images are not sent to a TabParcel OCR backend. An account requirement for Send to does not prevent local recognition, editing or export.
1. Choose an image or screenshot
Open Image OCR in the TabParcel extension and provide the image you want to recognize. If you are working from a screenshot, crop away unrelated areas so the printed text is easier to distinguish.
Low resolution, blur, decorative lettering, rotated text and complex layouts can reduce the result quality. A larger, sharper source is often more useful than repeatedly recognizing the same unclear image.
2. Set the recognition language
TabParcel recommends a recognition language from your interface language on first use and remembers your choice. You can change the OCR language independently from the extension interface.
- Simplified Chinese and Traditional Chinese
- English, Spanish and Portuguese
- Japanese and Korean
- German, French and Indonesian
Every non-English mode also supports English mixed with the selected language. Choose the language that matches the image rather than changing the interface language just for one document.
3. Run recognition and review the text
Start recognition, then check the editable result against the source. Pay particular attention to names, numbers, punctuation and line breaks before using the text in another document.
OCR can make mistakes even with a clear source. The current tool is primarily suited to horizontal printed text; it is not a promise of reliable handwriting or arbitrary document-layout recognition.
4. Save, copy or send the result
Edit the text, copy it, or choose a supported export such as TXT, Markdown or JSON. The tool also supports its ZIP export workflow. Downloaded files remain under your control.
To use Send to, sign into TabParcel with an active access period, open a supported destination folder and review what you are sending. Sending contacts that destination; it does not change local OCR into a server recognition service.
If the result is incomplete
Check the selected language, try a sharper source and crop to one clear text area. For a scanned PDF, first work with an image of the relevant page: copying the PDF’s text layer cannot recover text that exists only as pixels.
See OCR language questions or contact support with the extension version and reproduction steps. Avoid sending private document images unless you have deliberately chosen a safe example.