Emre Aksan

Hi, In our demo applications we used a bi-directional RNN classifier. You can find the relevant modules in [tf_train_hw_classification.py](https://github.com/emreaksan/deepwriting/blob/master/tf_train_hw_classification.py) and [tf_models_hw_classification.py](https://github.com/emreaksan/deepwriting/blob/master/tf_models_hw_classification.py). The content branch of our deepwriting model does classification,...

Hi, Now you can train the model. Please see the updated README. Unfortunately, our demo isn't online. You can first capture your handwriting samples by using a mobile device, preferably...

Hey, I will have a look and try to share at least the data conversion scripts. You can convert json data into numpy and train the model.

You can first call export PYTHONPATH=$PYTHONPATH:./source Without GMM, it would be similar to VRNN. Unfortunately, we don't have an unsupervised way of getting content.