Demo.TFLite

Demo and workshop to demonstrate the development flow for TFLite models.

Requirements

• Arduino IDE is preferred.
• PlatformIO is also possible, but hasn't been updated.
• Works on Windows and Linux
• See Python requirements in the other directories.

Content

The content is seperated in 3 main folders.

• arduinoide contains code to work with Tensorflow Lite through the Arduino IDE.
• MNIST contains the Python code to train and convert models using Tensorflow (Keras) and Tensorflow Lite.
• platformIO contains code specific for PlatformIO, which would allow to upload Arduino code through Visual Studio Code.

Converting TFLite models to C-Arrays

Read more on this in the MNIST > README file on the different ways to convert TFLite models to C-Arrays. In short: Use either Linux, or a Docker container.