Important question

#1

Hi Everyone,

I want to ask about the differences between machine language in Python and Edge, with details .please.

Thank you.
Bayan

#2

Hi @Bayan_khalid,

Python is a programming language that lets you program nearly anything in a high-level context. There are many machine learning packages (such as TensorFlow and Torch) that you can use with the Python language to train machine learning models.

Edge Impulse is a no- and low-code solution (i.e. an online platform) for developing machine learning models intended for embedded systems (e.g. microcontrollers and single board computers). The online Studio guides you through the process of collecting data, training a model, and deploying that model through the use of a graphical interface. Once you have a trained model, you can use other programming languages (such as C/C++ or Python) to perform inference with that model.

Hope that helps!

1 Like
#3

Hi shawn,
After I get my training model, I want to get a binary file that can run on the arrch64 hardware through cross-compilation. At this time, I need to use C++ to reason the neural network, right? How does this work work? Is there a demo for reference?
I have exported the .eim file of arrch64 according to our interface before, but my hardware does not have a software management package, so I cannot run our apt install EI, and can only use cross-compilation. How can I change it?

Thank you.

#4

Hi @y1165949971,

C++ is one option for creating a binary that performs inference with your model. You could also look into installing a package manager on your arrch64 system or installing the dependencies from source if you would like to use another language (e.g. Python).

#5

hello @shawn_edgeimpulse i will like to download the tensorflow lite models generated on the dashboard and use it as though i am using an originally trained tensorflow model on raspberry pi. I have tried but it is not working as i am getting segmentation errors. is it possible to use edge impulse models that way?

#6

Hi @Vic_fretz,

I’m confused by your question: are you trying to use the TFLite (.lite) model from the Edge Impulse Studio dashboard or one that you trained outside of Edge Impulse to perform inference?

Here is an example on how you might use a .lite file to perform inference with Python: TensorFlow Lite (TFLite) Python Inference Example with Quantization · GitHub. Note that you will need to copy in static features from your Edge Impulse project to the features array.