Moșnoi Ion

`jupyter --version jupyter core : 4.6.3 jupyter-notebook : 6.0.3 qtconsole : 4.7.3 ipython : 7.13.0 ipykernel : 5.2.1 jupyter client : 6.1.3 jupyter lab : 3.2.4 nbconvert : 5.6.1 ipywidgets...

hi, I am looking to do research in this area, can you provide the training data? I'd be interested in getting involved.

you need to change the learning rate, check if the images and labels are correct, change the stride parameter. the result from the readme are from multidimentional lstm.

vocabulary.txt contains the words in the dataset in order to use them for a language model(n-grams for example). the file is created in the utils.py. but if you have enough...

check out the `files = [data_pattern.format(j) for j in range(3)] if nameT=='train' else [data_pattern.format(j) for j in range(3,6)]` line from train.py to specify the number of training and testing tfrecords....

hi, i have used small images with 1-3 words with small letters. width-90 and height - 36 pixels. the best result was 12% for label rate error using overlaping windows...

first you get the images with lines by cropping the images from AMI dataset based on xml files. then use makeTFRecordFile() function from utils, the parameters of the function are...

it is because of the tensorflow version. i am using https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-0.12.0-cp27-none-linux_x86_64.whl if you need i can give you docker with tensorflow installed. also you could just delete the tf.cond and...

try https://github.com/floydhub/dl-docker

hi, i have tried to implement a multidimentional LSTM, here is a old version of my code http://stackoverflow.com/questions/42071074/multidimentional-lstm-tensorflow . I am not sure that it actually learns something. but we...