In the summary, merge i am getting the "TypeError: unhashable type: 'list' " error. Please Help me out

so i've running this command
python generate.py --noinfo --text="this was generated by computer" --bias=1.
in readme.md (https://github.com/Grzego/handwriting-generation/raw/master/imgs/example-1.PNG) - result of this command
how i can get only black letters or blue letters on image after running of this command?

Is it possible to get character level masks for the words generated?

Do you know how to changes the images data into xml files. I have the handwritting images data and labels, but I don't know how to change them into xml files.

Hey
What changes need to be made to enable training for other languages, other than the training data? I can see some style files being used, does that need to change as well? In which case, could you share how the style.npy files were created.

Hi! 👋

I'm following your instructions in README.md to get things running. For context: I'm running on a Microsoft Azure GPU, using Tensorflow 1.5 and Python 3.5. When I run python train.py, I'm getting the following error:

Traceback (most recent call last):
  File "/home/jameso2/handwriting-generation/.env/local/lib/python3.5/site-packages/tensorflow/python/pywrap_tensorflow.py", line 58, in <module>
    from tensorflow.python.pywrap_tensorflow_internal import *
  File "/home/jameso2/handwriting-generation/.env/local/lib/python3.5/site-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 28, in <module>
    _pywrap_tensorflow_internal = swig_import_helper()
  File "/home/jameso2/handwriting-generation/.env/local/lib/python3.5/site-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 24, in swig_import_helper
    _mod = imp.load_module('_pywrap_tensorflow_internal', fp, pathname, description)
  File "/home/jameso2/handwriting-generation/.env/lib/python3.5/imp.py", line 242, in load_module
    return load_dynamic(name, filename, file)
  File "/home/jameso2/handwriting-generation/.env/lib/python3.5/imp.py", line 342, in load_dynamic
    return _load(spec)
ImportError: libcublas.so.9.0: cannot open shared object file: No such file or directory

What could be causing this?

Thank you for this awesome work! 😊

Hi @Grzego,

I am trying to run the train.py (python train.py) file on the same IAM Online handwriting dataset. But getting some error (using tf 1.15)

Thank you for your help!!

私は
pip3 install tensorflow==13.1.2
で実行できた

def bivariate_normal(X, Y, sigmax=1.0, sigmay=1.0,
                     mux=0.0, muy=0.0, sigmaxy=0.0):
    """
    Bivariate Gaussian distribution for equal shape *X*, *Y*.
    See `bivariate normal
    <http://mathworld.wolfram.com/BivariateNormalDistribution.html>`_
    at mathworld.
    """
    Xmu = X-mux
    Ymu = Y-muy

    rho = sigmaxy/(sigmax*sigmay)
    z = Xmu**2/sigmax**2 + Ymu**2/sigmay**2 - 2*rho*Xmu*Ymu/(sigmax*sigmay)
    denom = 2*np.pi*sigmax*sigmay*np.sqrt(1-rho**2)
    return np.exp(-z/(2*(1-rho**2))) / denom

this function is depricated, just included it in the generate file and should do the work for you.

Greetings Sir
Thank a lot for providing such a nice implementation. However, I was facing some issues while implementing the code.
First of all in new numpy version allow_pickle is False by default so I solved one of the error by using allow_pickle=True in np.load statements in batch_generator.py

However, while trying to run train.py I am getting an error "NotImplementedError: Cannot convert a symbolic Tensor (model/rnn_1/while/rnn/rnn_model/window/add_1:0) to a numpy array."

This is same error I am getting while trying to use pretrained model

I am using Tensorflow 1.15.2 and numpy 1.19.5 and running on Google Colab

It's a great project,

Is the Arabic language included?

I am following the instructions in the README to generate handwriting. When I enter the command:
python generate.py --noinfo --text="this was generated by computer" --bias=1.

I get the following error:

Traceback (most recent call last):
File "generate.py", line 247, in
main()
File "generate.py", line 202, in main
fig, ax = plt.subplots(1, 1)
File "/home/jameso2/handwriting-generation/.env/local/lib/python3.5/site-packages/matplotlib/pyplot.py", line 1176, in subplots
fig = figure(**fig_kw)
File "/home/jameso2/handwriting-generation/.env/local/lib/python3.5/site-packages/matplotlib/pyplot.py", line 539, in figure
**kwargs)
File "/home/jameso2/handwriting-generation/.env/local/lib/python3.5/site-packages/matplotlib/backend_bases.py", line 171, in new_figure_manager
return cls.new_figure_manager_given_figure(num, fig)
File "/home/jameso2/handwriting-generation/.env/local/lib/python3.5/site-packages/matplotlib/backends/backend_tkagg.py", line 1049, in new_figure_manager_given_figure
window = Tk.Tk(className="matplotlib")
File "/usr/lib/python3.5/tkinter/init.py", line 1871, in init
self.tk = _tkinter.create(screenName, baseName, className, interactive, wantobjects, useTk, sync, use)
_tkinter.TclError: no display name and no $DISPLAY environment variable

For context, I am running the code on a Linux virtual machine with Ubuntu. My laptop is a MacBook.

Any idea how to resolve this? I'd appreciate any help!

Thank you for this great work :)

When I try to run the code, I am getting many errors because of new TensorFlow 2.x changes. So, I tried to use tf_upgrade_v2 script, but still, I get an error and not sure how to proceed,

More specifically: Substructure "type=LSTMStateTuple str=LSTMStateTuple(c=<tf.Tensor 'rnn/rnn_model/lstm_0/lstm_cell/add_1:0' shape=(64, 400) dtype=float32>, h=<tf.Tensor 'rnn/rnn_model/lstm_0/lstm_cell/mul_2:0' shape=(64, 400) dtype=float32>)" is a sequence, while substructure "type=ResourceVariable str=<tf.Variable 'model/rnn/Variable:0' shape=(64, 400) dtype=float32>" is not
Entire first structure:
[., (.,), [LSTMStateTuple(c=., h=.), LSTMStateTuple(c=., h=.), LSTMStateTuple(c=., h=.), ., ., .]]
Entire second structure:
[., (.,), [., ., ., ., ., ., ., ., .]]

It would be great if anyone give comments to reproduce this work in Tensorflow2.x

Would it be possible to output the text along a straight(er) line with this configuration?

There is some problem with the original web page，I can‘t log in to get the database。Could you send me a copy of it, please!
My E-mail is [email protected] .
Best wishes for you.

The data on this link: data/original-xml-part.tar.gz cannot be downloaded it's asking for username and password credentials. Is there another place where it can be found?

Hello,

I've been exploring different handwriting generators and would like to ask a question about this one.

May I please ask if your model, given the data you used to train it, is able to generate special symbols such as @, +, $ and £, ✓, as well as uppercase Q and Z?

Many thanks! :)

This is probably outside the scope of the "issues" but figure I'd ask.
I notice it doesn't take numbers. Is there away to add numbers to the xml data sets so it can also do numbers?

Hello,

First off, I would like to thank you for this amazing work. I was giving it a go hoping to directly use your pretrained model. However, I ran into the following error upon doing generate.py --text="this was generated by computer" --bias=1. Any help will be greatly appreciated.
I think it might be a renaming of some variables in the checkpoint file ?

NotFoundError: Key rnnlm/multi_rnn_cell/cell_10/basic_lstm_cell/bias/Adam_1 not found in checkpoint
	 [[Node: save/RestoreV2_17 = RestoreV2[dtypes=[DT_FLOAT], _device="/job:localhost/replica:0/task:0/cpu:0"](_arg_save/Const_0_0, save/RestoreV2_17/tensor_names, save/RestoreV2_17/shape_and_slices)]]

During handling of the above exception, another exception occurred:

NotFoundError                             Traceback (most recent call last)
/data.nfs/leo/handwriting-generation/generate.py in <module>()
    244 
    245 if __name__ == '__main__':
--> 246     main()

/data.nfs/leo/handwriting-generation/generate.py in main()
    132     with tf.Session(config=config) as sess:
    133         saver = tf.train.import_meta_graph(args.model_path + '.meta')
--> 134         saver.restore(sess, args.model_path)
    135 
    136         while True:

/data/dss-data-dir/code-envs/python/test_py3/lib/python3.6/site-packages/tensorflow/python/training/saver.py in restore(self, sess, save_path)
   1558     """
   1559     meta_graph.add_collection_def(meta_graph_def, key,
-> 1560                                   export_scope=export_scope)
   1561 

I am also running tensorflow 1.2.0

Hello, I am interested in training the model to other languages such as Spanish and Turkish but not sure how to generate the strokes sequence.Any help will be highly appreciated.
Thanks

After successfully implementing the training I find that starting at Epoch 18 the loss value is set to NAN and all the models created are of no use.

Is there any way to save all models such that the ones where loss is not NAN they can be used for generation.

Can the system be trained to imitate a SINGLE PERSON's handwriting?

Thanks.

Hello!

What's the license of this project?

Does anybody have the code converted to native Tensorflow 2 (without using tf.compat.v1) version?

Can I use my own style which is not included train dataset if I follow the format as you mentioned?

What are the input and output nodes of the model, these are required while i am converting the model to .tflite .

How to use it for Chinese/Korean/Japanese?

After preprocessing, I am facing "module 'tensorflow._api.v2.nn' has no attribute 'rnn_cell' " error. Please help me with it

I see no difference, irrespective of wheather option --animation is given or not.
Generative for pretrained model.
My python version is 3.4.2
My OS-X version is 1.13.6

the output form is slant, what's the reason of it? how to change it to horizontal?

It's quite a weird issue, not sure what might be causing it.

When I run the RNN to get the stroke data, it seems to add some garbage after my input string.

For example:

tsareva becomes Imgur

Here's how I'm using the model: TextRecognitionDataGenerator