OCR engine for all the languages

I am trying to build a Devanagari model using kraken. When I use default values for training it works but when I specify training and eval data separately, I get a codec error.

The following uses the same set of trainingdata.

This worked:

ketos train devatrain/*.png > devatrain.log

WARNING: Logging before flag parsing goes to stderr.
W0218 04:40:21.871934 127598883522176 __init__.py:74] TensorFlow version 1.15.0 detected. Last version known to be fully compatible is 1.14.0 .
Initializing model ✓

This gets the error:

ketos -v  train -t devatrain/*.png -e devatrain/*.png -o devatraintest > devatraintest.log

Traceback (most recent call last):
  File "/home/ubuntu/anaconda3/envs/py36/bin/ketos", line 8, in <module>
    sys.exit(cli())
  File "/home/ubuntu/anaconda3/envs/py36/lib/python3.6/site-packages/click/core.py", line 764, in __call__
    return self.main(*args, **kwargs)
  File "/home/ubuntu/anaconda3/envs/py36/lib/python3.6/site-packages/click/core.py", line 717, in main
    rv = self.invoke(ctx)
  File "/home/ubuntu/anaconda3/envs/py36/lib/python3.6/site-packages/click/core.py", line 1135, in invoke
    sub_ctx = cmd.make_context(cmd_name, args, parent=ctx)
  File "/home/ubuntu/anaconda3/envs/py36/lib/python3.6/site-packages/click/core.py", line 641, in make_context
    self.parse_args(ctx, args)
  File "/home/ubuntu/anaconda3/envs/py36/lib/python3.6/site-packages/click/core.py", line 940, in parse_args
    value, args = param.handle_parse_result(ctx, opts, args)
  File "/home/ubuntu/anaconda3/envs/py36/lib/python3.6/site-packages/click/core.py", line 1477, in handle_parse_result
    self.callback, ctx, self, value)
  File "/home/ubuntu/anaconda3/envs/py36/lib/python3.6/site-packages/click/core.py", line 96, in invoke_param_callback
    return callback(ctx, param, value)
  File "/home/ubuntu/anaconda3/envs/py36/lib/python3.6/site-packages/kraken/ketos.py", line 63, in _validate_manifests
    for entry in manifest.readlines():
  File "/home/ubuntu/anaconda3/envs/py36/lib/python3.6/codecs.py", line 321, in decode
    (result, consumed) = self._buffer_decode(data, self.errors, final)
UnicodeDecodeError: 'utf-8' codec can't decode byte 0x89 in position 0: invalid start byte

I'm trying to train Japanese OCR model using 2637 images with PAGE xml files. ketos train -d cuda:0 -f page -F 1 -q early --augment -o jpn *.xml

With 3.0b5 ketos consumes about 2.1GB memory and it is reasonable. With master(patch https://github.com/mittagessen/kraken/pull/199 is needed to run) ketos is eating more than 30GB memory and hangs.

It seems all images are stored in memory. Is there any option not to do that?

Hey there, I'd like to train a segmenter before leaving next week, and I understood you will release it soon. Any place where I can find the segmenter training format ?

@amitdo commented here on the specific RTL support in kraken. Since I am unsucessfully training OCR models for Hebrew with ocropy, I wonder if kraken could do the job. Can anyone introduce me to the details of kraken's RLT support? I could not find the related information in the documentation. Many thanks in advance!

I just installed kraken and kraken list gives only 10.5281/zenodo.2577813 (pytorch) - A generalized model for English printed text. Where are other models? I'm especially interested in Medieval Latin.

I ran the below command and receive an error as below.

This happens in both python3.6.8 and python3.7.2

 kraken -i bank.png bank.json segment  --remove_hlines --no-script-detect --scale 20 --pad 100 23 ocr --model en-default.pronn 
Loading RNN default	✓
Segmenting	✓
Traceback (most recent call last):
  File "/home/ram/code/lendsmart/py/kraken/venv/bin/kraken", line 10, in <module>
    sys.exit(cli())
  File "/home/ram/code/lendsmart/py/kraken/venv/lib/python3.6/site-packages/click/core.py", line 764, in __call__
    return self.main(*args, **kwargs)
  File "/home/ram/code/lendsmart/py/kraken/venv/lib/python3.6/site-packages/click/core.py", line 717, in main
    rv = self.invoke(ctx)
  File "/home/ram/code/lendsmart/py/kraken/venv/lib/python3.6/site-packages/click/core.py", line 1164, in invoke
    return _process_result(rv)
  File "/home/ram/code/lendsmart/py/kraken/venv/lib/python3.6/site-packages/click/core.py", line 1102, in _process_result
    **ctx.params)
  File "/home/ram/code/lendsmart/py/kraken/venv/lib/python3.6/site-packages/click/core.py", line 555, in invoke
    return callback(*args, **kwargs)
  File "/home/ram/code/lendsmart/py/kraken/venv/lib/python3.6/site-packages/kraken/kraken.py", line 220, in process_pipeline
    task(base_image=base_image, input=input, output=output)
  File "/home/ram/code/lendsmart/py/kraken/venv/lib/python3.6/site-packages/kraken/kraken.py", line 157, in recognizer
    for pred in bar:
  File "/home/ram/code/lendsmart/py/kraken/venv/lib/python3.6/site-packages/click/_termui_impl.py", line 285, in generator
    for rv in self.iter:
  File "/home/ram/code/lendsmart/py/kraken/venv/lib/python3.6/site-packages/kraken/rpred.py", line 301, in rpred
    preds = network.predict(line)
  File "/home/ram/code/lendsmart/py/kraken/venv/lib/python3.6/site-packages/kraken/lib/models.py", line 81, in predict
    o = self.forward(line)
  File "/home/ram/code/lendsmart/py/kraken/venv/lib/python3.6/site-packages/kraken/lib/models.py", line 69, in forward
    o = self.nn.nn(line)
  File "/home/ram/code/lendsmart/py/kraken/venv/lib/python3.6/site-packages/torch/nn/modules/module.py", line 489, in __call__
    result = self.forward(*input, **kwargs)
  File "/home/ram/code/lendsmart/py/kraken/venv/lib/python3.6/site-packages/torch/nn/modules/container.py", line 92, in forward
    input = module(input)
  File "/home/ram/code/lendsmart/py/kraken/venv/lib/python3.6/site-packages/torch/nn/modules/module.py", line 489, in __call__
    result = self.forward(*input, **kwargs)
  File "/home/ram/code/lendsmart/py/kraken/venv/lib/python3.6/site-packages/kraken/lib/layers.py", line 350, in forward
    o, _ = self.layer(inputs)
  File "/home/ram/code/lendsmart/py/kraken/venv/lib/python3.6/site-packages/torch/nn/modules/module.py", line 489, in __call__
    result = self.forward(*input, **kwargs)
TypeError: forward() missing 1 required positional argument: 'hidden'

Here is my pip list in venv

Let me know if i missed something.

Is my level of pips correct ?

Should i use a different level for torch ?

Hello, I tried to install kraken using pip3 and everything went fine, but I cannot use it. As soon as I try to get default, I have the following error message

$ kraken get default
Retrieving model	⣾Traceback (most recent call last):
  File "/usr/local/bin/kraken", line 11, in <module>
    sys.exit(cli())
  File "/usr/local/lib/python3.6/site-packages/click/core.py", line 722, in __call__
    return self.main(*args, **kwargs)
  File "/usr/local/lib/python3.6/site-packages/click/core.py", line 697, in main
    rv = self.invoke(ctx)
  File "/usr/local/lib/python3.6/site-packages/click/core.py", line 1092, in invoke
    rv.append(sub_ctx.command.invoke(sub_ctx))
  File "/usr/local/lib/python3.6/site-packages/click/core.py", line 895, in invoke
    return ctx.invoke(self.callback, **ctx.params)
  File "/usr/local/lib/python3.6/site-packages/click/core.py", line 535, in invoke
    return callback(*args, **kwargs)
  File "/usr/local/lib/python3.6/site-packages/click/decorators.py", line 17, in new_func
    return f(get_current_context(), *args, **kwargs)
  File "/usr/local/lib/python3.6/site-packages/kraken/kraken.py", line 346, in get
    partial(spin, 'Retrieving model'))
  File "/usr/local/lib/python3.6/site-packages/kraken/repo.py", line 46, in get_model
    raise KrakenRepoException(resp['message'])
kraken.lib.exceptions.KrakenRepoException: Bad credentials

I did remove kraken and tried with a pip2 install, but the result remains the same :

$ kraken get default
Retrieving model	⣾Traceback (most recent call last):
  File "/usr/local/bin/kraken", line 11, in <module>
    sys.exit(cli())
  File "/usr/local/lib/python2.7/site-packages/click/core.py", line 722, in __call__
    return self.main(*args, **kwargs)
  File "/usr/local/lib/python2.7/site-packages/click/core.py", line 697, in main
    rv = self.invoke(ctx)
  File "/usr/local/lib/python2.7/site-packages/click/core.py", line 1092, in invoke
    rv.append(sub_ctx.command.invoke(sub_ctx))
  File "/usr/local/lib/python2.7/site-packages/click/core.py", line 895, in invoke
    return ctx.invoke(self.callback, **ctx.params)
  File "/usr/local/lib/python2.7/site-packages/click/core.py", line 535, in invoke
    return callback(*args, **kwargs)
  File "/usr/local/lib/python2.7/site-packages/click/decorators.py", line 17, in new_func
    return f(get_current_context(), *args, **kwargs)
  File "/usr/local/lib/python2.7/site-packages/kraken/kraken.py", line 346, in get
    partial(spin, 'Retrieving model'))
  File "/usr/local/lib/python2.7/site-packages/kraken/repo.py", line 46, in get_model
    raise KrakenRepoException(resp['message'])
kraken.lib.exceptions.KrakenRepoException: Bad credentials

What did I miss ? (I use a macOS 10.12.6, python 2.7 or 3.6)

When I run the following command:

kraken -i OCR17plus/Data/Balzac1624_Lettres_btv1b86262420_corrected/png/Balzac1624_Lettres_btv1b86262420_corrected_0042.png results.txt segment -bl -i OCR17plus/Model/Segment/appenzeller.mlmodel ocr -m OCR17plus/Model/HTR/dentduchat.mlmodel

I have the following issue:

[37.7261] Failed processing OCR17plus/Data/Balzac1624_Lettres_btv1b86262420_corrected/png/Balzac1624_Lettres_btv1b86262420_corrected_0042.png: ndim 

The data used is available here: https://github.com/Heresta/OCR17plus

Any idea what the problem could be?

Hey there,

I can be wrong, but I have had weird behavior training the segmenter. I believe there is a mistake in the Click definition of the segmenter, but again, I can be wrong...

In the click, you got -bl/-tl:

https://github.com/mittagessen/kraken/blob/45f6bb0b1b1632de6077e6712f11d5461ddad63a/kraken/ketos.py#L158-L161

This value is of baseline (True or False) is passed down

https://github.com/mittagessen/kraken/blob/45f6bb0b1b1632de6077e6712f11d5461ddad63a/kraken/ketos.py#L260

to the topline kwarg and can be reused here:

https://github.com/mittagessen/kraken/blob/abe08ba4bec3778cf9e15c5cea7c0de503358a61/kraken/lib/segmentation.py#L440-L444

In this excerpt, it shows that topline=True will use topline (which makes sense).

If you create a file simply containing the same click declaration, such as :

import click

@click.command() 
@click.option('-bl/-tl', '--baseline/--topline', show_default=True, 
        default=False, help='Switch for the baseline location in the scripts. ' 
        'Set to topline if the data is annotated with a hanging baseline, as is ' 
        'common with Hebrew, Bengali, Devanagari, etc.') 
def bl(baseline): 
    print(baseline)

if __name__ == "__main__":
	bl()

you'll see that -bl gets makes baseline=True which will be directly used as topline=True. On the contrary, using -tl will set it to False. Basically, it seems to be the opposite of what is expected.

Hey, To follow up on https://github.com/mittagessen/kraken/issues/256, I felt like a new issue was in order, as the first was targeted at the inversion of -bl and -tl.

Before today, training and segmenting would provide a rather good segmentation but strike-through:

Since the commits of today, the segmentation is completely all over the place, using the same training set, eval set and command:

@gabays has seen the same issue

Hi there :) I was wondering if it'd be possible to boost a little the speed of building training set / valid set ? I looked at it and it seems quite "sequential"

https://github.com/mittagessen/kraken/blob/d39c45564df81d84bea58ee0067a48585a5f63e2/kraken/lib/train.py#L624-L632

I could take a chance at it if you do not have the time, and if you give me pointer on how you'd like it to be done (proxy function / reusing --threads, etc. )

mamba env create -f environment_cuda.yml --> leaves me with numpy 1.19.5 which in turn gives an error "RuntimeError: module compiled against API version 0xe but this version of numpy is 0xd" my attempts to update numpy failed as pip says : ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts. qudida 0.0.4 requires opencv-python-headless>=4.0.1, which is not installed. albumentations 1.3.0 requires opencv-python-headless>=4.1.1, which is not installed. kraken 4.2.1.dev84 requires numpy<=1.23.0, but you have numpy 1.24.1 which is incompatible. coremltools 4.1 requires numpy<1.20,>=1.14.5, but you have numpy 1.24.1 which is incompatible.

Is there any possibility to run it in Windows? Can Ubuntu be installed from Windows Store and then be used?

Does this model "Kraken:arabPersPrBigMixed_best(1)" ring a bell? If yes, from where to download it? More details in this video in minute 05:21 in the Open ITI project

Thanks,

Medo Hamdani

WARNING Non-encodable sequence ︎◻ ךו... encountered. Advancing one code point. codec.py:131 etc

this is repeated for every epoch

suggestion: suppress this as the difference of training and testing codec already has a separate yellow warning before launch of training.

The training was started with at least 200 epochs and 20 tries to get a better model:

ketos train -f page -t list.train -e list.eval -o Juristische_Konsilien_Tuebingen+256 -d cuda:0 --augment --workers 24 -r 0.0001 -B 1 --min-epochs 200 --lag 20 -w 0 -s '[256,64,0,1 Cr4,2,8,4,2 Cr4,2,32,1,1 Mp4,2,4,2 Cr3,3,64,1,1 Mp1,2,1,2 S1(1x0)1,3 Lbx256 Do0.5 Lbx256 Do0.5 Lbx256 Do0.5 Cr255,1,85,1,1]'

Early stopping would have stopped after stage 111, but training continues because at least 200 was requested. Instead of producing stage 112, 113, 114, ..., it stays at stage 112 and repeats the validation step again and again:

stage 109/∞ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 7366/7366 0:00:00 0:05:14 val_accuracy: 0.87676  early_stopping: 18/20 0.87974
stage 110/∞ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 7366/7366 0:00:00 0:05:18 val_accuracy: 0.87542  early_stopping: 19/20 0.87974
stage 111/∞ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 7366/7366 0:00:00 0:05:18 val_accuracy: 0.87760  early_stopping: 20/20 0.87974
stage 112/∞ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 0/7366 -:--:-- 0:00:00  early_stopping: 20/20 0.87974Trainer was signaled to stop but the required `min_epochs=200` or `min_steps=None` has not been met. Training will continue...
stage 112/∞ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 11802/7366 0:00:00 0:08:02 val_accuracy: 0.87345  early_stopping: 20/20 0.87974
Validation  ━━━━━━━━━━╸━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 223/826    0:00:40 0:00:16                                                     

Hey, I just discovered that when you fine tune a model, the new model keeps the logs from the base model, such that it has the dev score from the first model in model.user_defined_metadata["kraken_meta"])["accuracy"]

Is it intended ? :)