Image to text (OCR)
Get the words out of a photo, screenshot or scanned page, ready to copy. The image is read on this device and never uploaded.
- Nothing is uploaded
- Works offline
- Open source
- No limits, no watermark
Done
Open the network tab. It stays empty.
Add files and run the tool to see every request this tab makes.
Your files are processed on this device and never sent to a server. Reload the page to clear everything from memory.
How it works
- Drop one or more images (JPG, PNG, HEIC, WebP, screenshots or scans), or tap to pick them.
- Keep line breaks as they are in the image, or join them into paragraphs for pasting into a document or an email.
- Read text. The text appears on the page with a Copy button, and each image can be saved as a .txt file.
In your browser, not on a server
Your browser decodes the image (libheif or the Squoosh decoders step in for HEIC, AVIF and JPEG XL), enlarges small screenshots so letters are big enough to read, and hands the pixels to Tesseract, the open-source OCR engine first built at HP and later developed by Google, compiled to WebAssembly by tesseract.js (Apache-2.0). Its neural network model for English is served by this site and loaded the first time you run it. The text is recognised line by line inside a worker in your tab.
Compared with upload-based tools: The pictures people want text from are often private: a letter, a receipt, a bank statement, a prescription, a screenshot of a conversation. Online OCR sites upload the image to their server to read it and keep it for hours, and several limit free use or ask for an account. Here the image never leaves your device, and the page keeps working with Wi-Fi off.
Verify it yourself. Load this page, turn off Wi-Fi or switch to airplane mode, and run the tool. Or open your browser's developer tools (F12, or Cmd-Option-I on a Mac), pick the Network tab, and watch it while you convert. The panel above does the same count for you.
Questions
Is my image uploaded?
No. The text is recognised by Tesseract, the open-source OCR engine, compiled to WebAssembly and running inside this page. The engine and its English model (about 6 MB) are downloaded from this site the first time you use it, then the tool works offline. Switch off Wi-Fi after the first run and it still reads images.
How accurate is it?
Very good on printed text that is sharp and straight: screenshots, documents, scans, signs and receipts photographed head-on. It struggles with handwriting, curved or angled text, heavy shadows and decorative fonts. For a phone photo, fill the frame with the text and hold the phone parallel to the page. Always check numbers you plan to rely on.
Which languages does it read?
English, including accented Latin letters in names and places. More languages are on the way; each needs its own model of a few megabytes.
Can it read a scanned PDF?
Not directly yet. Turn the pages into images with PDF to JPG first, then drop them here. If the PDF already has selectable text, PDF to Word gets it out faster and keeps the paragraphs.
Why is the first run slower?
The first time, your browser downloads the OCR engine and model from this site. They are cached, so later runs start straight away, even offline. Reading a full page takes a few seconds on a laptop and a little longer on a phone.
More image tools
HEIC to JPG
Turn iPhone and iPad photos into JPG or PNG files that open anywhere. Your photos never leave this device.
Image Converter
Drop any image, see what it really is, pick what you need. Converted on your device.
Compress & Resize Images
Make photos small enough to email or upload. Batch resize, set a quality, see the savings before you download.
Remove EXIF Data
See what a photo reveals, then delete it. Lossless: pixels are untouched, only the metadata is removed.