Which model is good for document images (scanned documents, administrative documents, papers) ?

[Philo-theMoonLover]

I tested on CNNIQA, MUSIQ, DBCNN, UNIQUE, PIQE, BRISQUE. With my dataset, I found that they work quite well with face images, but the scores for document images are usually much lower and it is difficult to differentiate good and bad images.

Can you suggest a suitable model for document images? This is not issue, thanks for your help.

[chaofengc]

Unfortunately, there isn't a model specifically designed for document images, and there is also a lack of datasets tailored to this task.