TensorFlow.jl with Google's Coral board

[ghost]

Hello

I want to use TensorFlow.jl to work with Google’s Coral board (https://coral.ai/docs/edgetpu/models-intro/).
In order to do this, I have to convert the .pb graph to a quantized .tflite graph. So:

1. Is there an API in TensorFlow.jl to convert from TensorFlow to TensorFlow Lite?

FIGURE 1

2. Or is it possible to use the Python API for doing this? (https://www.tensorflow.org/lite/convert/python_api)
   Check Figure 2. I got it from the JuliaCon presentation. It was said that it is possible to share code with Python users.

FIGURE 2

[oxinabox]

Here is docs on how to export the graph

https://malmaud.github.io/TensorFlow.jl/latest/saving.html#Saving-and-restoring-models-1

So you should be able to give that graph to the tflite converter, in theory.