OpenCV text detection

Question

I have a system that extracts text from pictures ((MS word, tables, ..) documents with text more than 2 lines for example), but I do not want pictures that obviously have no text in it (such as textures, photos, ...). I've tried algorithm's from detect-text-in-images-with-opencv and detect-text-region-in-image-using-opencv, but all algorithms works not so good for example with these images

Answer 1

OpenCV has a module containing text detection.

https://docs.opencv.org/master/da/d56/group__text__detect.html

the module comes with sample code and documentation.

there is also a method based on deep learning called "EAST". here's the sample:

https://github.com/opencv/opencv/blob/master/samples/dnn/text_detection.cpp