In the example, 25 images are generated. How do I generate one image

https://ww2.mathworks.cn/help/deeplearning/ug/train-generative-adversarial-network.html?searchHighlight=gan&s_tid=srchtitle_support_results_1_gan

In this GAN example, 25 images were generated, I wanted to generate 1 image, and I changed the code

‘numValidationImages = 25’ to ‘numValidationImages = 1’

This is the image I generated after training. What went wrong

Answer 1

[Mike Croucher]

The imtile command should be removed when you only have one image:

Go down to the Generate New Images section and change numObservations

numObservations = 1;
ZNew = randn(numLatentInputs,numObservations,"single");
ZNew = dlarray(ZNew,"CB");

if canUseGPU
    ZNew = gpuArray(ZNew);
end

XGeneratedNew = predict(netG,ZNew);

% I = imtile(extractdata(XGeneratedNew));       % Use this if numObservations is > 1
I = extractdata(XGeneratedNew);                 % Use this if numObservations == 1
I = rescale(I);
figure
image(I)
axis off
title("Generated Images")

[bohan]

Thank you very much. I have solved it.Your advice is very useful

Answer 2

[Aradhy]

numObservations = 1;

ZNew = randn(numLatentInputs,numObservations,"single");

ZNew = dlarray(ZNew,"CB");

if canUseGPU

ZNew = gpuArray(ZNew);

end

XGeneratedNew = predict(netG,ZNew);

% I = imtile(extractdata(XGeneratedNew)); % Use this if numObservations is > 1

I = extractdata(XGeneratedNew); % Use this if numObservations == 1

I = rescale(I);

figure

image(I)

axis off

title("Generated Images")

[Image Analyst]

Your answer is just a copy of @Mike Croucher's answer. Did you mean to do that?