Deployed model shows wrong classification

[davidib19]

Question/Issue: Hello, my spectrogram classification model works fine on the Edge Impulse web page. However, when I download the c++ library and build it, or when I deploy it on my arduino 33 BLE sense it won’t work. It prints the wrong classification. I had this same problem months ago and you fixed it, maybe the fix got deleted on an update?

Project ID: 66887

Context/Use case:

Thank you for your help
Regards,
David Córdova

[mikeh9999]

I have the same problem with image detection. Did you figure out a solution?

[aurel]

Hi @davidib19,

As you use 3 spectrogram blocks it may be linked to how you fill the features array. Could you share your main code that includes how you use this buffer?

Thanks,

Aurelien

[davidib19]

Hi @aurel
I used 3 spectrogram blocks as suggested by @louis in this topic:

ERR: Failed to run DSP process (-1004) spectrogram

I tried using the example stand alone inference master to use my model in my Windows PC but i get this error. ERR: Failed to run DSP process (-1004) run_classifier returned: -5 Begin output [0.00000, 0.00000] End output The model classifies using the spectrogram block How can I fix this? I appreciate the help

I don’t have the main code right now but I have the same problem with the built firmware for the arduino 33ble sense

Thanks,
David