Drop an image and click Extract text. The OCR engine runs 100% on your device: your image is never uploaded. On first use the engine (about 7 MB) downloads once and is cached by your browser.
Pull editable text out of photos, scans and screenshots — free, in your browser.
Drop an image and click Extract text. The OCR engine runs 100% on your device: your image is never uploaded. On first use the engine (about 7 MB) downloads once and is cached by your browser.
Extract editable text from photos, scans and screenshots using OCR that runs entirely in your browser. No upload, no sign-up: drop an image, click Extract text, then copy or download the result as a text file.
A phone photo of a printed page usually extracts in a few seconds. A 2,000 × 1,500 screenshot of a settings screen produces clean, selectable text that can be pasted into notes, an email or a document.
OCR (optical character recognition) is performed by the Tesseract recognition engine compiled to WebAssembly, running locally in your browser with English language data. It analyses the shapes of letters in the image and reconstructs words and lines. The image is processed in device memory only — it is never sent to a server.
No. The OCR engine runs in your browser using WebAssembly. Your image never leaves your device, which makes this safe for documents, IDs and private screenshots.
The first use downloads the recognition engine and English language data (about 7 MB in total) and your browser caches them. Later extractions on this device skip that download.
Straight-on, well-lit images where the text is large and in focus. Crop away clutter, avoid photographing screens at an angle, and prefer scans or screenshots over camera photos where you can.
Handwriting is unreliable with this engine, and this version includes English data only. Printed English text — including mixed case, numbers and common punctuation — works well.
Low resolution, blur, low contrast or text over busy backgrounds confuse character shapes. Try a larger, sharper version of the image, or crop it to just the text area and run it again.