How it works
Upload a scanned PDF or a photo of a document. Each page is rendered as an image and run through an on-device OCR engine that recognizes characters directly from the pixels — the same underlying technique used by paid OCR services, just running locally in your browser instead of on their servers.
A note on the first run
The OCR engine (a few megabytes of code and language data) loads the first time you use this tool in a session. Your browser caches it, so it won't need to reload on repeat use in the same session.