OCR algorithm - distinguishes between a text image and an object image

Question

I am writing a program that extracts content from the logo of different sites. I use OCR to extract text from a logo, but I want to optimize the program and want to use OCR only on logos that have text, but I don’t know how to determine if the logo contains text or not? any method

Answer 1

This is the case when we need to know if there is text in it. It is different from OCR.

The algorithm that is considered the best today is the conversion of the stroke width. It was developed by Ephstein under Microsoft in 2010. It does not use any machine learning goals.

You can get more detailed information from this article: Detecting text in natural scenes with stroke width conversion

Or watch a video about it.

There is an implementation of this algorithm here .