Unable to perform assignment because the size of the left side is 1-by-4096 and the size of the right side is 6-by-6-by-256

imds = imageDatastore('archive', 'IncludeSubfolders',true, ...

'LabelSource','foldernames');

[imdsTrain,imdsValidation] = splitEachLabel(imds,0.7,'randomized');

net = alexnet;

inputSize = net.Layers(1).InputSize

layersTransfer = net.Layers(1:end-3);

numClasses = numel(categories(imdsTrain.Labels))

layers = [

layersTransfer

fullyConnectedLayer(numClasses,'WeightLearnRateFactor',20,'BiasLearnRateFactor',20)

softmaxLayer

classificationLayer];

pixelRange = [-30 30];

imageAugmenter = imageDataAugmenter( ...

'RandXReflection',true, ...

'RandXTranslation',pixelRange, ...

'RandYTranslation',pixelRange);

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

'DataAugmentation',imageAugmenter);

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

options = trainingOptions('sgdm', ...

'MiniBatchSize',10, ...

'MaxEpochs',6, ...

'InitialLearnRate',1e-4, ...

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

'ValidationData',augimdsValidation, ...

'ValidationFrequency',3, ...

'Verbose',false, ...

'Plots','training-progress');

netTransfer = trainNetwork(augimdsTrain,layers,options);

[YPred,scores] = classify(netTransfer,augimdsValidation);

YValidation = imdsValidation.Labels;

accuracy = mean(YPred == YValidation)

numValidation = numel(imds.Files);

featuresValidation = zeros(numValidation, 4096); % 4096 is the output size of 'pool5' layer in AlexNet

for i = 1:numValidation

img = readimage(imds, i);

img = imresize(img, netTransfer.Layers(1).InputSize(1:2));

features = activations(netTransfer, img, 'pool5');

featuresValidation(i, :) = features;

end

[Petrov Daniel]

Hello,

I try to train an image database using AlexNet and I have to do a clusterisation, but I get this error while trying to extract the information from the last layer before fully connected layers. Can anyone give advice how to solve this, please!

[Stephen23]

@Petrov Daniel: please show us the complete error message. This means all of the red text.

Answer 1

[Yash]

Hi Petrov,

I understand that you are facing issues while extracting the information from the "pool5" layer of the "AlexNet" network.

According to the MATLAB documentation of the "AlexNet" network, the output size of the "pool5" layer in AlexNet is 6*6*256. The figure below also depicts the same:

The above figure is taken from the MATLAB documentation available at the following link: https://www.mathworks.com/help/deeplearning/ref/alexnet.html

Hence, the assumption that the output size of the "pool5" layer in AlexNet is 4096 is incorrect, and this size mismatch is causing the error you are facing.

If you want a 1D array of features, you can use the "reshape" function. However, the size would still remain 9216 (6*6*256). The documentation of the reshape function is available at the following link: https://www.mathworks.com/help/matlab/ref/reshape.html

Hope this helps!