Test or validate a model using (same number of images form each subfolder)

Hi, i have three subfolders (good, moderate and severe) with images(3608, 406 and 200 images respectivly)

1-the accuracy is (98.18), is it ok?

2-after training my model,how to take (100 images from each subfolder to test (validate) the model) ?

3-what is the code for confusion matrix

Thanks

my code is next:

imds = imageDatastore('C:\Users\Rayan\Desktop\9_8_balance_data\R_9_1_GSM', ...

'IncludeSubfolders',true, ...

'LabelSource','foldernames');

[imdsTrain,imdsValidation] = splitEachLabel(imds,0.85,'randomized');

numTrainImages = numel(imdsTrain.Labels);

idx = randperm(numTrainImages,16);

net = resnet50;

inputSize = net.Layers(1).InputSize;

lgraph = layerGraph(net);

% call the find layers to Replace at : edit(fullfile(matlabroot,'examples','nnet','main','findLayersToReplace.m'))

edit(fullfile(matlabroot,'examples','nnet','main','findLayersToReplace.m'))

[learnableLayer,classLayer] = findLayersToReplace(lgraph);

[learnableLayer,classLayer] %#ok<NOPTS>

numClasses = numel(categories(imdsTrain.Labels));

if isa(learnableLayer,'nnet.cnn.layer.FullyConnectedLayer')

newLearnableLayer = fullyConnectedLayer(numClasses, ...

'Name','new_fc', ...

'WeightLearnRateFactor',10, ...

'BiasLearnRateFactor',10);

elseif isa(learnableLayer,'nnet.cnn.layer.Convolution2DLayer')

newLearnableLayer = convolution2dLayer(1,numClasses, ...

'Name','new_conv', ...

'WeightLearnRateFactor',10, ...

'BiasLearnRateFactor',10);

end

lgraph = replaceLayer(lgraph,learnableLayer.Name,newLearnableLayer);

newClassLayer = classificationLayer('Name','new_classoutput');

lgraph = replaceLayer(lgraph,classLayer.Name,newClassLayer);

layers = lgraph.Layers;

connections = lgraph.Connections;

layers(1:20) = freezeWeights(layers(1:20));

lgraph = createLgraphUsingConnections(layers,connections);

augimdsTrain = augmentedImageDatastore(inputSize(1:2),imdsTrain)

augimdsValidation = augmentedImageDatastore(inputSize(1:2),imdsValidation);

miniBatchSize=10;

valFrequency = floor(numel(augimdsTrain.Files)/miniBatchSize);

options = trainingOptions('sgdm', ...

'MiniBatchSize',10, ...

'MaxEpochs',10, ...

'InitialLearnRate',0.0007, ...

'Shuffle','every-epoch', ...

'ValidationFrequency',valFrequency, ...

'ValidationData',augimdsValidation, ...

'Verbose',false, ...

'Plots','training-progress');

net = trainNetwork(augimdsTrain,lgraph,options);

[YPred,probs] = classify(net,augimdsValidation);

accuracy = mean(YPred == imdsValidation.Labels);

idx = randperm(numel(imdsValidation.Files),100);

R=1;

for j =1:25

figure(j)

for i = 1:4

subplot(2,2,i)

I = readimage(imdsValidation,idx(R));

imshow(I)

label = YPred(idx(R));

title(string(label) + ", " + num2str(100*max(probs(idx(R),:)),3) + "%");

R=R+1;

end

end

Answer 1

[Image Analyst]

Regarding accuracy. It might be. But let's say you were building a dog and cat trainer and gave it 10000 images of dogs and 30 images of cats. Then, once trained, you give it the test set comprised of both dogs and cats - 9818 dog images and 172 cat images - and it called every single one a dog. That's 98.18 accurate. Is it good enough? It didn't find a single cat but it was 98.18% accurate.