features extraction for images with different features size in a matrix error?

I am trying to extract features from a dataset of images the problem that the features extraction algorithm gives different features size for each image, for example, the first image 1000 features and the second 2000. and it gives me an error because of that when I try to store them in a matrix. what should I do??

imds=imageDatastore('E:\dataset test','IncludeSubFolders',true,'LabelSource','foldernames');
trainingFeatures=[];

[imds_5k, imds_extra] = splitEachLabel(imds,15);
trainingLabels=imds_5k.Labels;  
for i = 1:numel(imds_5k.Files)         % Read images using a for loop
 img = readimage(imds_5k,i);
   trainingFeatures(i,:) = example(img);
   
end

this is the error: Unable to perform assignment because the size of the left side is 1-by-86386 and the size of the right side is 1-by-73638.

Answer 1

[Ameer Hamza]

You can use a cell array

imds=imageDatastore('E:\dataset test','IncludeSubFolders',true,'LabelSource','foldernames');
[imds_5k, imds_extra] = splitEachLabel(imds,15);
trainingLabels=imds_5k.Labels;
trainingFeatures=cell(1,numel(imds_5k.Files));
for i = 1:numel(imds_5k.Files)         % Read images using a for loop
    img = readimage(imds_5k,i);
    trainingFeatures{i} = example(img);  
end

[yasmin almashagbah]

thank you for your answer, now I can not use the data or export it as an Xls file it will export it as 1*81465, not the real data and I can not convert it into a matrix since each row have a different length??

[Ameer Hamza]

Following code will append 0 to make all vectors of equal lengths and the create a matrix which can be written to xls file

n_max = max(cellfun(@numel, trainingFeatures));
trainingFeatures = cellfun(@(x) {[x; zeros(n_max-numel(x),1)]}, trainingFeatures);
trainingFeatures = [trainingFeatures{:}];
writematrix(trainingFeatures, 'file.xls');

[yasmin almashagbah]

it gives this error

Dimensions of arrays being concatenated are not consistent.

Error in

Untitled3>@(x){[x;zeros(n_max-numel(x),1)]}

[Ameer Hamza]

Oh! I forgot that example() returns row vector. Try the following code

n_max = max(cellfun(@numel, trainingFeatures));
trainingFeatures = cellfun(@(x) {[x zeros(1,n_max-numel(x))]}, trainingFeatures);
trainingFeatures = vertcat(trainingFeatures{:});
writematrix(trainingFeatures, 'file.xls');

[yasmin almashagbah]

it works, thank you

[yasmin almashagbah]

if you can help me!!

https://www.mathworks.com/matlabcentral/answers/687524-index-in-position-1-is-invalid-error-after-running-multiple-times?s_tid=prof_contriblnk