All in-browser AI tools

Extract text from an image

Choose a screenshot, a photo of a document or a scan, and the Florence-2 model reads the text from it. Everything happens in your browser.

The file is processed on your device and is not uploaded to a server.

Model: Florence-2 base · license MIT · 275 MB

Works best with Latin-script text; Cyrillic is recognized poorly.

Questions

Which languages does it read?
Florence-2 reads Latin-script text best: English, Spanish, Serbian Latin and similar. Cyrillic and Asian scripts are recognized poorly.
Why is the model so large?
It understands the whole picture, not just letters: about 300 MB. It is downloaded once and then starts from the browser cache.
Is the image uploaded?
No. Recognition happens on your device.

Related DataHunter tools

Other AI tools that run in your browser