PII Detector
Scan text for emails, phones, and card-like patterns locally.
OpenPull English text out of a screenshot or scan, with recognition running on your own device.
Pick a screenshot, scan or photo with text.
The engine loads once, then reads the image locally.
Recognized text appears in the output panel.
Select the text and proofread numbers and names.
Someone sent an error message as a screenshot, or you have a scanned letter and need to quote two paragraphs from it. Retyping is slow and introduces typos. Image OCR reads the text out of the image for you.
Choose an image file and click Run OCR. On the first run the page downloads the Tesseract.js engine and its English language data, which takes a moment; after that, recognition starts sooner. The recognized text appears in the output panel, or a note if no text was found, and you can select it to copy. One image is processed per run, and recognition uses the English model, so other languages and scripts come out poorly.
Results depend heavily on the image. Crisp screenshots of interface text usually come out close to perfect. Phone photos of paper work better if the page is flat, evenly lit and fills the frame. Cropping to the area you need and increasing contrast before you run OCR often helps more than anything else. Proofread numbers, since 0 and O or 1 and l are common mix-ups.
The image is passed to the OCR engine inside your browser and is not uploaded; only the engine files are downloaded. If the extracted text might contain personal data, scan it for emails and phone numbers before you paste it elsewhere. Photos can also carry location data, which the EXIF stripper removes. More local tools are on the privacy tools page.
Keyboard shortcut
Answers for this tool. For site-wide help, open the FAQ hub.
English only. The tool loads the English language model, so text in other languages may come out with missing accents or garbled characters, and non-Latin scripts will not be recognized.
Only neat block capitals, and even then results are unreliable. Tesseract is designed for printed text, so expect many errors on cursive or fast handwriting.
The OCR engine and English language data are downloaded the first time you click Run OCR. Later runs in the same session reuse what has already loaded.