why a Gray image is shown as a Colored image on CNN deep learning ?

CNN network of deep learning reads my gray image as a colored image. Whenever, tried to change the diminsions to gray [ 227 227 1], the system gives me error

layers = [

imageInputLayer([227 227 3],"Name","data")

convolution2dLayer([11 11],94,"Name","conv1","BiasLearnRateFactor",2,"Stride",[4 4])

reluLayer("Name","relu1")

crossChannelNormalizationLayer(5,"Name","norm1","K",1)

maxPooling2dLayer([3 3],"Name","pool1","Stride",[2 2])

groupedConvolution2dLayer([5 5],94,2,"Name","conv2","BiasLearnRateFactor",2,"Padding",[2 2 2 2])

Answer 1

[Image Analyst]

That's right. For most predefined network architectures, they were built to handle color images. Just make your gray scale images into color images and don't worry about it. The network will eventually learn during training that it doesn't need to use the other two color channels.

[Mohamed Elbeialy]

I do not want to convert the gray image to color. I am looking to use the gray image as is and let the network read it and change the diminsions to gray [ 227 227 1]. Is that possible ?

[Image Analyst]

I think so but you'd have to build your network from scratch using the network designer, rather than use a pre-built network like AlexNet, ResNet, GoogLeNet, etc.

[Mohamed Elbeialy]

Unfortunately, this not possible because I already used the pre-built network and am stuck now with the dimensions point. Any other suggestions?

[Image Analyst]

Why can't you just convert to color? What's wrong with doing that?

[Mohamed Elbeialy]

I hope so, but this type of image can not be converted to color. It's HeatMapSum image

[Image Analyst]

It can be converted:

rgbImage = cat(3, grayImage, grayImage, grayImage);

or

rgbImage = ind2rgb(grayImage, gray(256));

[Mohamed Elbeialy]

I tried but it did not work. If you can do, go ahead and show me the result

[Image Analyst]

WHAT did not work? The cat() function? Or your training/classification/prediction process?

I can't really download all your training images. Sorry. I suggest you call tech support and ask them to walk you through it step by step.