Inference with neural net trained with AI-Studio

[sallogy]

Hello,

I exported  the neural Net Coffee_or_Not trained with AI Studio in a .config file (+source files).  How this can be used in a BSEC 2.0 library environment for inference, eg. detecting gases the model was trained for?  Can you provide some examples/documentation?

Thanks,

Gyorgy

[BSTRobin]

Hello sallogy ,

Currently, the config file was only used for BME688 Development Kit.
After you exported the neural Net Coffee_or_Not trained with AI Studio in a .config file. Based on config file, do you mean you would like to used it as standalone application in BSEC?

[sallogy]

Hello,

yes,  I mean exactly this - how the neural net trained with AI Studio and exported as a config file can be used with the BSEC library (or some other software) to do inference. For example,  to run on a microcontoller  in an air purification device, do gas detection periodically  and to send a signal to initiate air purification when a certain type of gas is detected.

Thanks,

György

[BSTRobin]

Hello sallogy,

In "BME AI-Studio document", it supported to export the algorithm as BSEC configuration file. Refer BME688 integration guide document(under "integration_guide" folder as you have downloaded BSEC package) for how to use it in your project.
https://www.bosch-sensortec.com/media/boschsensortec/downloads/application_notes_1/bst-bme688-an001....