Google trials optical character recognition for Google Now on Tap

by: Robert TriggsMarch 8, 2016


Google appears to still be tinkering with the ins and outs of its Google Now on Tap feature. A number of users have begun noting that the software is now capable of performing optical character recognition (OCR) using your regular camera viewfinder or from a previously taken picture.

After having read the text from an image, the system uses Google Now to bring up a number of related search options based on contextual information from the picture. Results for people, brands, books, and news all appear to be displayed. Navigation information can also crops up from time to time, and Now on Tap is also able to pull phone numbers from an image for you to call back later.

Google Now on Tap OCR

Google Now on Tap optical character recognition examples, courtesy of Reddit users.

It seems to work in a similar way to Google Translate’s OCR for directly translating text from images. Interestingly though, the feature isn’t working with general image recognition, it has to be text based. The OCR feature still looks to be in its early stages though. Text read through the viewfinder doesn’t always come through completely intact, and different phones appear to be reproducing varying results from the same source material. Similarly, using different gallery apps to look at pictures produces hit and miss results.

If you want to test out what’s going on with Google Now on Tap’s optical character recognition feature so far, you’ll obviously need to have Android 6.0 Marshmallow installed for a start. It’s not clear how widely available this feature is just yet, but try tapping on some text in your gallery app and see what happens.

Google Now on Tap reviewSee also: Google Now on Tap gets new features, now supports French, Spanish and four other languages3
  • Rob

    Isn’t this basically Google Glass, except using your phone’s camera?

    • scoupdogg

      nope,you can use it like ANYWHERE on a website, on pre-downloaded photos, on literally anything which might have optical character.

  • Javid Nazim Mammadov

    Nice, this is actually pretty useful