I trained the model through spectral analysis. After deploying it to the chip, I found that only the following metrics were displayed: Timing: DSP 11 ms, inference 0 ms, anomaly 0 ms. Predictions:
heart_breath: 0.99609
heart_nobreath: 0.00000 Regarding this kind of data, I want to know how to view the strongest value within the frequency domain of the FFT?
If you’d like to see the intermediate FFT feature output you can use the debug flag when calling your classifier. This varies depending on your target, for example if you’re building a C++ library you can change the third argument from false to true here to get the full debug output: https://github.com/edgeimpulse/example-standalone-inferencing/blob/31b09a58417d0a1aa62d5372155967e86d476b2a/source/main.cpp#L46C1-L47C51