An error " Subscripted assignment dimension mismatch" on line ' features(i​,:)=extrac​tHOGFeatur​es(img,'Ce​llSize',ce​llsize);' what might be the problem for the given code?

clc;
close all;
clear all;
DatabaseDir=fullfile(toolboxdir('vision'),'elephants','dataset');
TestDir=fullfile(toolboxdir('vision'),'elephants','testset');

          Dataset=imageSet(DatabaseDir,'recursive');
          Testset=imageSet(TestDir,'recursive');

%dataset images 

figure;
subplot(1,2,1);
imshow(Dataset(1).ImageLocation{1});
subplot(1,2,2);
imshow(Dataset(1).ImageLocation{2});

%testset images

figure;
subplot(1,2,1);
imshow(Testset(1).ImageLocation{1});
subplot(1,2,2);
imshow(Testset(1).ImageLocation{2});

img = read(Dataset(1), 4);
[hog_4x4, vis4x4] = extractHOGFeatures(img,'CellSize',[4 4]);
cellsize = [4 4];
hogFeatureSize = length(hog_4x4);

trainFeatures=[];
trainLabels=[];
l=numel(Dataset);
for i=1:l
    numimages=Dataset(i).Count;
    features=zeros(numimages,hogFeatureSize,'single');
    for j=1:numimages
        img=read(Dataset(i),j);
        level=graythresh(img);
        img=im2bw(img,level);

          features(i,:)=extractHOGFeatures(img,'CellSize',cellsize);

      end;
      labels = repmat(trainingSet(digit).Description, numimages, 1);
      trainFeatures=[trainingFeaturres;features];
      trainLabels=[trainingLabels;labels];

end
% fitcecoc uses SVM learners and a 'One-vs-One' encoding scheme.
classifier = fitcecoc(trainFeatures, trainLabels);

Answer 1

[Walter Roberson]

That code relies upon all the images in the dataset collection being nearly exactly the same size -- mod(size(TheImage),4) has to be the same for all of them. If that is not the case then the number of HOG features is going to be different for some of the images.

A common way to handle this is to imresize() to the same size for all of the images.