Working now
Read text from an image
Point this at a scanned page, a photo of a document or a screenshot and it returns the text as something you can copy and edit. The recognition engine runs inside your browser, so the picture is never sent anywhere.
How to use it
- Drop your scan, photo or screenshot.
- Wait for the engine to download the first time. It is about 7 MB and your browser keeps it after that.
- Read the text below the file, then copy it or save it as a .txt file.
Questions
Why is the first run slow?
The recognition engine and the English model have to download once, roughly 7 MB together. Your browser stores them, so every run after that starts immediately.
How accurate is it?
On a clean scan or a screenshot of typed text, close to perfect. On a phone photo taken at an angle, or on faint print, it drops. Handwriting is not supported at all.
Which languages does it read?
English only for now. Each extra language is a separate download, so they are being added one at a time rather than all at once.
Can it read a PDF?
Not yet. Export the page as an image first, or wait for the PDF version of this tool.
What does the layout setting do?
It tells the engine what to expect. A page of text suits documents. Single line suits a photo of one line, like a serial number. Scattered words suits signs and labels.