deep network designer, how do I test a trained network against a new dataset? I am able to see the predictions for all the images?

Hi,

So first of all I am not a computer scientist, just starting to use MatLab.

I have tried deep network designer to classify between 3 classes. Saved the script, the results etc. Now I want to use this trained network on a new dataset and also I really need to see the results for each imagine in the dataset. Is this possible and is anyone able to help?

Thanks a lot

Andreea

Answer 1

[David Willingham]

Hi Andreea,

Firstly, glad to hear your starting your journey of deep learning in MATLAB!

Your code will likely look like this:

imds = imageDatastore('MyImageFolder', ...

'IncludeSubfolders',true, ...

'LabelSource','foldernames');

[YPred,scores] = classify(mytrainedNet,imds);

idx = randperm(numel(imds.Files),4);

figure

for i = 1:4

subplot(2,2,i)

I = readimage(imds,idx(i));

imshow(I)

label = YPred(idx(i));

title(string(label));

end

I would also recommend studying this example. It shows the suggested workflow for performing transfer learning :

https://www.mathworks.com/help/deeplearning/ug/transfer-learning-using-pretrained-network.html