Implementing fast OCR application on android device

asked 2016-10-18 04:52:02 -0600

Greetings. We're evaluating the feasibility of an android application with OpenCV, which should work as follows:

  • 1) The user clicks a button in the app to start the image acquisition
  • 2) A video feedback coming from the phone camera should be displayed on the screen (possibly not full screen)
  • 3) The user will frame an object which is painted with an alphanumeric string. Often this string of text is curved. As an example:

image description

  • 4) Without explicitly taking a photo, the application should be able to "read" the image stream coming from the camera (which, as stated, is also shown in a box inside the application) and recognize the alphanumeric string in less than one second, and with 90%+ accuracy (ideally 99%+)
  • 5) The recognized string is displayed as text in a textbox

Given the requirements (less than 1 second to identify the text and 90%+ accuracy), is the computational power of a mid-range android device enough to fulfill them with ease?

Thanks for your kind help!

edit retag flag offensive close merge delete