Pre-trained model + single language [Solved]

[khawar-islam]

Dear doctr members,

I am recognizing handwritten documents which contains Hangul characters+Some English characters+digits+special signs like ({},(), % etc). It is very hard to collect english characters+digits+special signs like ({},(), % etc) data. Is pretrained model have these capabilities? If yes, I will use the pretrained model and train on korean characters.

OR

I have to train a model from scratch by collecting all data english characters+digits+special signs like ({},(), % etc)

[khawar-islam]

I have trained a master model with single characters dataset and used pre-trained model to recognize korean characters but it did not work out.