Extract the readable text from a photo, screenshot, or scanned image using optical character recognition (OCR).
Accuracy depends on image clarity — clean, high-contrast, well-lit text recognizes best; blurry or handwritten text is less reliable.
The current OCR engine is tuned for English text.
The image itself stays in your browser and is recognized on your device. The first run does download the open-source OCR engine and its English language data from a public CDN, which is program data — not your image.
The OCR engine and language model are fetched once, then cached by your browser. After that, recognition starts much faster.
Shoot or crop tightly around the text, keep the page flat and evenly lit, avoid shadows and glare, and hold the camera square to the page. High-contrast printed text recognizes far better than handwriting or low-resolution screenshots.
This tool takes image files. For a scanned PDF, convert the pages to images first with PDF to JPEG, then run those images through OCR here.