Hi EI team,
Currently, I can run a NN model for image classification in my MCU. But the inputs of the model is raw features and I can only get it from edge impulse. In addititon, I want to display my input images, also.
So, I think that I must do both steps (running infernece and displaying image) locally without using EI.
Could you please help guide me how to convert from an image into raw features since I did not see any relating documnents, presently?
The conversion of an image into raw features typically involves preprocessing steps like resizing, normalization, and extracting relevant features using techniques such as edge detection or color histograms. This does happen on our servers and the raw features are extracted automatically during data processing.
If you want to work on the ML side see our python sdk to get you going in a traditional ML dev space:
If you are new to the area you can start here:
See also the following thread: What are the features extracted from the images when using linear regression?
Hopefully this will get you started,
Best
Eoin