Question/Issue:
Hello dear Edge Impulse community,
A few weeks ago, I got curious and purchased the new Arduino Q, featuring a Qualcomm processor—but this device is truly driving me up the wall.
I want to use it to perform object detection; I initially attempted this using Ultralytics with OpenCV, but that proved simply too resource-intensive for such a small board.
I then switched to the Edge Impulse workflow using App Lab—a workflow that is, after all, designed specifically for Arduino by default. So, I linked my Arduino to Edge Impulse, trained my model, and it appeared in the Object Detection Brick exactly as it should.
I then copied the example project, loaded and selected my model, but the same thing happens every time: it consistently defaults to using the standard model that comes pre-loaded on the Arduino as an example. The app.yaml file correctly references my specific project, yet the standard example—featuring the dog, cat, keyboard, etc.—is the one that always loads.
I am slowly starting to lose my mind over this. I’ve already tried modifying the app.yaml file via VS Code and moving the model files around in the directories, but nothing seems to work. Despite following the official Arduino workflow, I simply cannot get my custom model to run.
For this reason, I am now turning to the forum for help. Can anyone here tell me how to resolve this issue?
Thank you all very much, and best regards!
Actual Outcome:
[wont work]
Reproducibility:
-
[X] Always
-
[ ] Sometimes
-
[ ] Rarely
-
Custom Blocks / Impulse Configuration: [object detection]