handwriting-ocr
handwriting-ocr copied to clipboard
words extraction is not happening
I cannot find words-final folder, & I am unable to extract data.
I see that I forgot to include further details on how to extract the data. I guess that you already downloaded the data. Placed them according to the instructions. Then you have to go into src/data/
folder and run following instructions:
python data_extractor.py -d all
python data_normalization.py -d all
python data_create_sets.py --csv -d all
You can all also specify exact datasets for processing by replacing all with their names. For example:
python data_extractor.py -d iam cvl
There might still be some issues, so let me know if you encounter some other problem.
"import enchant error" as enchant is not available for 64 bit OS so what can be other option to install enchant
What did you try to install? You should run: pip install pyenchant
.
I just found out, that it's no longer under maintenance. I will try to replace it.
Getting error while executing OCR.ipynb file with MODEL_LOC = '../models/word-clas/' + LANG + '/CTC/Classifier2' but when i executed with MODEL_LOC = '../models/char-clas/' + LANG + '/CharClassifier' it got executed without any error.
I made this change also "CHARACTER_MODEL = Model(MODEL_LOC,'word_prediction')" instead of CHARACTER_MODEL = Model(MODEL_LOC)
"TypeError: Cannot interpret feed_dict key as Tensor: The name 'x:0' refers to a Tensor which does not exist. The operation, 'x', does not exist in the graph."
Yes, it is because CTC model requires little bit different inputs. OCR notebook right now works only with char classifier. Take a look into ocr_evaluator.ipynb
at class WordCycler()
. I hope you will be able to figure out neccesseray changes.
Once I have some time, I will extend the OCR notebook to support all models.