The value of 'ValidationData' is invalid. Duplicate table variable name: 'input'. error during neural network training

I want to train an autoencoder that copies the input image.

I used augmentedImageDatastore for image resizing, then combined two augmentedImageDatastore to use as input and responses using the 'combine' function for the autoencoder.

However, If i run the code below, I see the error message:

The value of 'ValidationData' is invalid. Duplicate table variable name: 'input'.

I checked the 'combine' function for 'ImageDatastore' work well but I am not sure why it is not work for 'augmentedImageDatastore'.

Thank you for your help.

digitDatasetPath = fullfile("my path");
imds = imageDatastore(digitDatasetPath, ...
    IncludeSubfolders=true,LabelSource="foldernames");

imds.ReadSize = 1024;
imds = shuffle(imds);
[imdsTrain,imdsVal,imdsTest] = splitEachLabel(imds,0.95,0.025);

% Create augmented datastores for input and response
augmentedTrainInput = augmentedImageDatastore([224,224,3], imdsTrain);
augmentedTrainOutput = augmentedImageDatastore([224,224,3], imdsTrain);

augmentedValInput = augmentedImageDatastore([224,224,3], imdsVal);
augmentedValOutput = augmentedImageDatastore([224,224,3], imdsVal);

augmentedTestInput = augmentedImageDatastore([224,224,3], imdsTest);
augmentedTestOutput = augmentedImageDatastore([224,224,3], imdsTest);

% Combine the augmented datastores for input and response
dsTrain = combine(augmentedTrainInput, augmentedTrainOutput);
dsVal = combine(augmentedValInput, augmentedValOutput);
dsTest = combine(augmentedTestInput, augmentedTestOutput);

% Define the network layers
imageLayer = imageInputLayer([224,224,3]);

encodingLayers = [ ...
    convolution2dLayer(3,8,Padding="same"), ...
    reluLayer, ...
    maxPooling2dLayer(2,Padding="same",Stride=2), ...
    convolution2dLayer(3,16,Padding="same"), ...
    reluLayer, ...
    maxPooling2dLayer(2,Padding="same",Stride=2), ...
    convolution2dLayer(3,32,Padding="same"), ...
    reluLayer, ...
    maxPooling2dLayer(2,Padding="same",Stride=2)];

decodingLayers = [ ...
    transposedConv2dLayer(2,32,Stride=2), ...
    reluLayer, ...
    transposedConv2dLayer(2,16,Stride=2), ...
    reluLayer, ...
    transposedConv2dLayer(2,8,Stride=2), ...
    reluLayer, ...
    convolution2dLayer(1,3,Padding="same"), ...
    clippedReluLayer(1.0), ...
    regressionLayer];    

layers = [imageLayer,encodingLayers,decodingLayers];

options = trainingOptions("adam", ...
    MaxEpochs=50, ...
    MiniBatchSize=imds.ReadSize, ...
    ValidationData=dsVal, ...
    ValidationPatience=5, ...
    Plots="training-progress", ...
    OutputNetwork="best-validation-loss", ...
    ExecutionEnvironment="multi-gpu", ...
    DispatchInBackground=true,...
    Verbose=true);

net = trainNetwork(dsTrain,layers,options);
modelDateTime = string(datetime("now",Format="yyyy-MM-dd-HH-mm-ss"));
save("trainedImageToImageRegressionNet-"+modelDateTime+".mat","net");    

ypred = predict(net,dsTest);
testBatch = preview(dsTest);

idx = 1;
y = ypred(:,:,:,idx);
x = testBatch{idx,1};
ref = testBatch{idx,2};
montage({x,y});

Answer 1

[Joss Knight]

augmentedImageDatastore returns a table so cannot be trivially combined. You should first transform it to convert it into a cell array of just the input values.

>> amds = augmentedImageDatastore([224,224,3],digitDatastore);
>> read(amds)

ans =

  128×2 table

         input         response
    _______________    ________

    {224×224 uint8}       0    
    {224×224 uint8}       0    
    {224×224 uint8}       0    
    {224×224 uint8}       0    
    {224×224 uint8}       0    
    {224×224 uint8}       0    
    {224×224 uint8}       0    
    {224×224 uint8}       0    

           :              :     

	Display all 128 rows.

>> amds = transform(augmentedImageDatastore([224,224,3],digitDatastore), @(t)t.input);
>> read(amds)

ans =

  128×1 cell array

    {224×224 uint8}
    {224×224 uint8}
    {224×224 uint8}
    {224×224 uint8}
    {224×224 uint8}
    {224×224 uint8}
    {224×224 uint8}
    {224×224 uint8}
    {224×224 uint8}
    {224×224 uint8}
    {224×224 uint8}
    
           :