cifar_dataset.m

%% Download and Load CIFAR-10 Data
clear;
url = 'https://www.cs.toronto.edu/~kriz/cifar-10-matlab.tar.gz';
downloadFolder = tempdir;
filename = fullfile(downloadFolder,'cifar-10-matlab.tar.gz');

dataFolder = fullfile(downloadFolder,'cifar-10-batches-mat');
if ~exist(dataFolder,'dir')
    fprintf("Downloading CIFAR-10 dataset (175 MB)... ");
    websave(filename,url);
    untar(filename,downloadFolder);
    fprintf("Done.\n")
end

%% Load CIFAR-10 Data
[XTrain,YTrain,XValidation,YValidation] = loadCIFARData(downloadFolder);

trainLabels = YTrain;
testLabels = YValidation;

trainData = zeros(size(XTrain,1),size(XTrain, 2), 1, size(XTrain, 4));
testData = zeros(size(XValidation,1),size(XValidation, 2), 1, size(XValidation, 4));

%% Convert to Grayscale
% convert training images to grayscale
for i = 1:size(XTrain, 4)
    trainData(:,:,1,i) = rgb2gray(XTrain(:,:,:,i));
end

% convert test images to grayscale
for i = 1:size(XValidation, 4)
    testData(:,:,1,i) = rgb2gray(XValidation(:,:,:,i));
end


%% Reshape the Data
trainData = squeeze(trainData);
testData = squeeze(testData);
trainData = orzReshape(trainData, 1);
testData = orzReshape(testData, 1);


%% Convert to numerical labels
testLabels = convertToNumericalLabels(testLabels);
trainLabels = convertToNumericalLabels(trainLabels);

%% Convert to Matlab Subspace Format
trainData = convertDatasetToMatlabSubspaceFormat(trainData, trainLabels);
testData = convertDatasetToMatlabSubspaceFormat(testData, testLabels);

%% Display images
imageSize = 32;
data = reshape(trainData, imageSize, imageSize, size(trainData, 2), size(trainData, 3));
figure; % Create a new figure window
for i = 1:10 % Loop through each of the 10 classes
    % random integer, used to display random image
    idx = randi([1, 100]);
    img = squeeze(data(:,:,idx,i)); % Extract the first 16x16 image of the i-th class
    subplot(2, 5, i); % Arrange the plots in a 2x5 grid
    imagesc(img); % Display the image
    axis square; % Make each subplot square in shape
    colormap gray; % Use grayscale colors
    title(sprintf('Class %d', i)); % Title each subplot with its class number
end

%% Save the dataset
save('cifar10.mat', 'trainData', 'testData');