is it possible to train model to get Text Readability of image from "Image Quality Assessment" ?

[harishkhollam]

The problem I am trying to solve is find "Text Readability of an Image" I am using Tesseract to extract text from images, but I am not able to understand what are the parameters the tesseract is considering for a good/bad text extraction.

In order to analyse the image I am using "Image Quality Assessment" to get a image quality technical score.

The models does not focus on how good the text is in the image i.e Readability , can I train this model on the images I have and if yes how ?

[clennan]

Hi, yes, I guess you could train a model on text readability in images. You would need a labeled dataset with many examples of varying degrees of readability and their corresponding scores. Then you could train a model as described in our documentation.

Let me know if you have further questions