This goal of this tutorial is to describe, how to build a fast prototype for a REST API to serve an experimental deep learning system for digital ink utilizing Wacom's Universal Ink Model.
In order to train a machine learning model, you can pick your favorite framework:
- Keras - https://keras.io/
- Tensorflow - https://www.tensorflow.org/
- PyTorch - https://pytorch.org/
- Scikit learn - https://scikit-learn.org/stable/
- ...
The project model-trainer contains a sample the extract sensor data.
The projects contains a folder models where you should store the trained models.
FastAPI is an easy-to-use framework to build Python REST APIs (micro-services).
The sample service is packaged as a Docker container, thus it can be deployed in several cloud infrastructures, for instance:
- Amazon's AWS
- Microsoft Azure
- Digital Ocean
- Heroku
Exemplary, we will use Heroku deploy.
1. Step: Create a Heroku account and configure your backend
2. Step: Log into your container registry.
$ heroku container login
3. Step: Build docker image and push it to Heroku registry.
$ heroku container:push web --app <APP-NAME>
3. Step: Build docker image and push it to Heroku registry.
$ heroku container:release web --app <APP-NAME>
Access logging: Build docker image and push it to Heroku registry.
$ heroku logs --tail --app <APP-NAME>
You can find more samples on how to use Wacom's Universal Ink Library here
For further details on using the SDK see WILL SDK for ink documentation
The API Reference is available directly in the downloaded SDK.
If you experience issues with the technology components, please see related FAQs
For further support file a ticket in our Developer Support Portal described here: Request Support
Join our developer community:
This sample code is licensed under the MIT License