Overview / Usage

Developing deep learning models is entirely different from the stage where it is to be taken to production and made to use. In this project, I show how to develop a simple image classification model with the FashionMNIST dataset from the Zalando Research team using TensorFlow 2.0. I then show to save the model weights and load it in a Python script and how one can use Flask to wrap this in a REST API endpoint.

The REST API then needs to be deployed somewhere so that a developer can consume it for his use. I enlist the steps to deploy the REST API on the Heroku platform.

Methodology / Approach

The plan is to first develop a model with which we can proceed. After that, comes the step of saving the model weights for later use.

After the model weights are saved, we need to use it to make inferences on the new images but in a production setting. To do this, we need a script that would allows us to run it in a standalone manner.

Technologies Used

Frameworks used:

• TensorFlow

• Flask

• NumPy

Repository

https://github.com/sayakpaul/GDE-Collaboration