Prototype (sample) text recognition from video app for Android

I need a proof-of-concept piece of code of sample quality - just good enough to show that this idea can work. It needs to run on Android phones and do the following: - Start the camera in video mode and show the live video on the screen. - Watch for any blocks of text in the video image. - When text is seen, draw a box around it in the video image - Even as the camera moves, maintain the box accurately around the text Requirements: - Draw the box within 3 seconds of the text first appearing in the field of view - Handle at least 10 simultaneous blocks of text - Keep the boxes aligned on the text with less than 0.5 seconds lag after camera motion It is not needed to perform full OCR on the text. It is enough just to recognize that part of the image is a block of text. It is ok to use open-source AR or OCR libraries to assist in this project. Deliverables: - Installable .apk file - Full source code - Build instructions