Invalid input data. Invalid size of channel dimension. Layer expects input with channel dimension size 3 but received input with size 1.

i used alexnet model for custom training. I loaded rgb images as input but i recieve an error that the input channel size is 1. please help me to resolve this error. i have attached the code snippet.

imds = imageDatastore('N:\oraldata',...

'IncludeSubfolders',true,...

'labelSource','foldernames',...

'FileExtensions','.jpg');

imdrgb=imds.read;

XTrain = imdrgb;

YTrain = imds.Labels;

classes = categories(YTrain);

numClasses = 2;

net=alexnet;

dlnet = dlnetwork(lgraph)

velocity = [];

numEpochs = 2;

miniBatchSize = 30;

numObservations = numel(YTrain);

numIterationsPerEpoch = floor(numObservations./miniBatchSize);

initialLearnRate = 1e-4;

momentum = 0.9;

decay = 0.01;

% % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % %

iteration = 0;

start = tic;

% Loop over epochs.

for epoch = 1:numEpochs

% Shuffle data.

idx = randperm(numel(YTrain));

XTrain = XTrain(idx);

YTrain = YTrain(idx);

% Loop over mini-batches.

for i = 1:numIterationsPerEpoch

iteration = iteration + 1;

% Read mini-batch of data and convert the labels to dummy

% variables.

idx = (i-1)*miniBatchSize+1:i*miniBatchSize;

X = XTrain(idx);

A=char(X);

Y = zeros(numClasses, miniBatchSize, 'single');

for c = 1:numClasses

Y(c,YTrain(idx)==classes(c)) = 1;

end

% Convert mini-batch of data to dlarray.

dlX = dlarray(single(A),'SSCB');

% Evaluate the model gradients and loss using dlfeval and the

% modelGradients function.

[gradients,loss] = dlfeval(@modelGradients,dlnet,dlX,Y);

% Determine learning rate for time-based decay learning rate schedule.

% learnRate = initialLearnRate/(1 + decay*iteration);

% Update the network parameters using the SGDM optimizer.

[dlnet.Learnables, velocity] = sgdmupdate(dlnet.Learnables, gradients, velocity, learnRate, momentum);

% Display the training progress.

if plots == "training-progress"

D = duration(0,0,toc(start),'Format','hh:mm:ss');

addpoints(lineLossTrain,iteration,double(gather(extractdata(loss))))

title("Epoch: " + epoch + ", Elapsed: " + string(D))

drawnow

end

end

end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

function [gradients,loss] = modelGradients(dlnet,dlX,Y)

dlYPred = forward(dlnet,dlX);

dlYPred = softmax(dlYPred);

loss = crossentropy(dlYPred,Y);

gradients = dlgradient(loss,dlnet.Learnables);

end

and i am gettin error as,

Layer 'data': Invalid input data. Invalid size of channel dimension. Layer expects input with channel dimension size 3

but received input with size 1.

[M J]

Hey, did you find a solution to your problem? I'm having a similar issue.

Cheers.

[drummer]

up

Answer 1

[Snehal]

Hello @lakshmipriya premkumar,

The error message indicates that the model ‘alexnet’ expects the input with 3 channels, but your code is providing data with only 1 channel. A possible reason for this could be that the input image you’re using is in grayscale format instead of RGB.

You can check the format of the input image and convert it to RGB (if they’re not already) using the following approach:

function image_rgb = convertToRGB(image) 
    % Check if the image is grayscale 
    if size(image, 3) == 1 
        % Duplicate the single channel to create a 3-channel image 
        image_rgb = cat(3, image, image, image); 
    else 
        % If already RGB, no conversion needed 
        image_rgb = image; 
    end 
end 

% Invoking the above defined method 
img = convertToRGB(img); 

Below are some documentation links for your reference:

• https://www.mathworks.com/help/matlab/ref/double.cat.html
• https://www.mathworks.com/help/matlab/ref/double.size.html#d126e1653855:~:text=szdim%20%3D%20size(A%2Cdim)%20returns,the%201%2Dby%2D2%20row%20vector%20szdim.

Hope this helps!