Image to Text (OCR)
Upload a photo of a document, sign, or screenshot to recognize and extract its text using on-device optical character recognition (Tesseract) — the image is never uploaded to a server, and recognition quality depends on image clarity and language.
Your images are processed entirely on your device by an in-browser OCR engine — they're never uploaded to a server.
Optical character recognition (OCR) results are approximate and not guaranteed to be 100% accurate, especially with low-quality images or unclear fonts. Review the extracted text before relying on it.
Drag images containing text here, or choose them
You can select more than one image (JPG, PNG, or WebP)
What this tool does
Extracts text from images using optical character recognition in a language you choose, with batch processing and downloadable results.
When to use it
- A screenshot containing text you want to copy.
- A photo of a book page or document whose text you need.
- A sign or notice you photographed and want in writing.
- An image containing data you would rather not retype.
How to use it
- Put your images on the drop area or choose them.
- Choose the language of the text in the image — the single most important setting here.
- Wait; on the first run the OCR engine loads and that can take a few seconds.
- View the text, download it as a text file, or use "Download all as ZIP".
A practical example
Input
A screenshot of a clearly rendered message.
Output
The message's text in editable, copyable form.
Limitations
- Accuracy depends entirely on image quality: good lighting, a straight-on shot and crisp text give excellent results, while a blurry, skewed or low-contrast image gives poor ones.
- Choosing the wrong language ruins the result even on an excellent image.
- Handwritten and decorative lettering is far harder than printed text.
- Formatting is not preserved: columns and tables can come out interleaved.
- Connected scripts such as Arabic are harder for OCR engines than Latin, so expect to proofread the output.
Privacy and processing
Recognition runs entirely inside your browser and no image is uploaded — which matters, because the images people transcribe are usually documents or screenshots containing personal data.
Technical notes
- The engine is Tesseract running in the browser, with every asset and language file served from this same site rather than a third-party network.
- It loads only when an OCR tool is actually opened, so its several megabytes never reach other pages.