This repository contains the sourcecode related to my blogpost series on Medium on deep learning model exploration, translation and deployment using the EMNIST dataset.

1. Clone the repository and change to the folder
2. If you are using conda, just create a new environment from env.yml with conda env create -f env.yml. This will create the environment emnist_dl.
3. Activate the environment with conda activate emnist_dl
4. Install the package with python setup.py install
5. Work through the JuPyter notebooks provided in notebooks/

I provide six JuPyter notebooks to guide through the code. I advise you to set everything up, read each blogpost and go trough the referenced notebooks to try things out yourself.

Part 1:

• 1_exploration_data_EMNIST.ipynb
• 2_exploration_model.ipynb
• 3_translation_ONNX_GraphPipe.ipynb

Part 2:

• 4_production_TFServing.ipynb
• 5_production_Webserver.ipynb
• 6_conclusion_Serving_Performance_Comparison.ipynb

The code has been tested with Docker Version 18.06.1-ce-mac73 (26764)

This project has been set up using PyScaffold 3.1rc2. For details and usage information on PyScaffold see https://pyscaffold.org/.