Am I missing something obvious in trying to upload an impulse build via sketch to my Nicla Vision?

Question/Issue: I have created a basic model that classifies two objects. I want to upload this to my Nicla Vision. Please note that I am new to Arduino’s and Edge Impulse, so I may be missing something obvious here. I have concluded that I am best to use Arduino Library build as I would like to send my results to an IoT dashboard, so I will need to edit the code in the sketch. I have built Arduino Library Deployment. I then followed the steps in the Arduino IDE "Sketch, Include Library, add .Zip library’. I was able to do this successfully. Then I went to 'File, Examples ‘the name of my Edge Impulse project’. I was assuming here would be a sketch of the model I have made, but all I see is example sketches for the nicla sense, nicla vision, portenta h7 etc. The steps online show for the one listed ‘static buffer’ (Arduino library - Edge Impulse Documentation), however this looks like another example?

Can anyone provide me clarity on the situation, please?

Project ID: 220863

Context/Use case: Image Classification/Object detection

Hello @LaurenMcGarry,

If you want to run your model on your Nicla Vision, you can select the Examples->Your imported library->nicla-vision->nicla-vision-camera:

Then you can just compile and deploy that model, check that it’s working (you can open the serial console and see the output logs).

Once that is verified, you can start using that example to create a custom application.




Hi @louis,

I have been using this successfully, but now I have tried to create an anomaly detection using the accelerometer on the board. I created the model, and then Build the Arduino Library. However, when I import the library, there is only the camera, fusion, microphone and microphone continious options. Is this model I trained supposed to be in one of these?



Hello @LaurenMcGarry,

Just select the sensor fusion one. After the run_impulse function, there is a condition to also print the anomaly score if you have one anomaly detection block in your project.



1 Like

Thank you, this works!