OCR Application

I’m trying to build an OCR using Google Vision. Right now I’ve built it partially and what it does is it scans and recognize everything that it finds on the screen.

I only want a certain portion of the screen to be scanned like a rectangular portion on the screen instead of the whole screen.

Actual Screen

What I want

In the second Image, I want only the non-blurred area to act as an OCR.

How can I achieve this?