handwriting-ocr
handwriting-ocr copied to clipboard
How to Train 'models/gap-clas/CNN-CG'
I don't see any file to train this model 'models/gap-clas/CNN-CG' (Since this is the model used most of the files for character separation) Also I don't see any file to create gap data for character Can you please give me some additional info to create gap data for character/words
Training of CNN-CG
model (it is bit inconsistent with naming). Is done in GapClassifier.ipynb
. But that isn't the best model for separating letters in word. I would recommend you to use GapClassifier-BiRNN.ipynb
which gives better results.
Gap data for this model can be created using script WordClassDM.py
. This script displays each word from given location and by clicking you specify positions where should be the word separated into individual letters.
yes @Breta01 we followed what you have mentioned above My target is to train multiple handwriting , If I use single handwriting I get good accuracy , but when goes more than one handwriting Accuracy decreases, Can you say where I can concentrate so that Accuracy improves ?
That's not easy to say. Best way is to take some sample text (for which the algorithm fails) and evaluate individual steps. Ideally, you should evaluate each step as if the previous steps worked perfectly. That way you can determine where you should put the most effort.
I can't see the GapClassifier-BiRNN.ipynb and GapClassifier.ipynb. maybe names changed ?