Can I avoid image normalization in deep learning

When I use trainNetwork() to run a deep learning job on a bunch of images. The second comment I'm reading after execution is that my images are being normalized:

Initializing image normalization.

Can I switch this off somewhere?

[Greg Heath]

Yes, but I don't recommend it Greg

[Mads]

Well how?

[Brian Derstine]

https://www.mathworks.com/help/deeplearning/ref/nnet.cnn.layer.imageinputlayer.html#d120e48025

Answer 1

[Raphael Ruschel]

You can switch it off by declaring your input layer as follows imageInputLayer([size], 'Normalization','none')