How to modify spectrograms so that there is no effect of amplitude on their classification?

Hello

Is there any way to bring different classes of spectrograms to comparable amplitude levels so that when they are used for classification, the deep learning algorithm focuses on other aspects (like the presence or absence of harmonics)?

Thanks in advance.

Answer 1

[Image Analyst]

If you want, you can try rescale

[NASRIN AKTER]

I was following this example:

https://www.mathworks.com/help/radar/ug/pedestrian-and-bicyclist-classification-using-deep-learning.html

Here they log-transformed the stfts and then normalized the amplitude to [0,1]. Does that mean the algorithm is already using information other than amplitude to classify?

[Image Analyst]

Yes, it looks more at the shape of the signal (spectrogram image) than the amplitude.