Amit Dovev

Please remove any mention of `joining_thread` from the C++ Core Guidelines.

>LITIS/IMPACT Workshop : **Recent** Development in OCR for Digital Libraries It's from March 2011. Not so recent... :)

On the other hand, it's quite recent if you compare it to #26...

> It looks like the ocropy-model has more parameters/weights in the LSTM layers than the clstm-model: They share wci, wgi, wgf, wgo, but the ocropus model has wip, wfp, wop...

> The line normalization and preprocessing is the same for both types of models. From ocropy README.md > ## CLSTM vs OCRopy > > .... > > Python and C++...

@melody-rain, You have a typo in your profile page. It should be 'Computer **vision** and deep learning'.

Thanks for reporting. I don't think it's a good solution to read all the dataset in each loading.

~~The best solution IMHO would be to encode all the \~128.000 unicode codepoints at the first load.~~ Update: It's not a good idea, see comments below.

I don't suggest to actually do training on all those chars...

In the meantime, don't use your temporary solution. I believe it will mess your model.