Creating an imdb structure

Hi,

I have written a new createIMDB function following an online tutorial on Deep Learning by German Ros. The original imdb was written for a binary classification problem. I attempted to change it for a three class problem. Each of my class 'cat', 'dog' and 'rat' contain say, three images, for simplicity. My function goes like this:

function imdb = createIMDB_RAM1(folder)

% imdb is a matlab struct with several fields, such as:
 % - images: contains data, labels, ids dataset mean, etc.
 % - meta: contains meta info useful for statistics and visualization
 % - any other you want to add

imdb = struct();

% let's assume we have a folder with three
% subfolders "cat" "dog" and "rat" containing images
% for a multi-class problem

positives = dir([folder '/cat/*.jpg']);

negatives = dir([folder '/dog/*.jpg']);

neutral = dir([folder '/rat/*.jpg']);

imref = imread([folder '/cat/', positives(1).name]);

[H, W, CH] = size(imref);

% number of images

NPos = numel(positives);

NNeg = numel(negatives);

NNeu = numel(neutral);

N = NPos + NNeg + NNeu;

% we can initialize part of the structures already

meta.sets = {'train', 'val'};

meta.classes = {'cat', 'dog', 'rat'};

% images go here

images.data = zeros(H, W, CH, N, 'single');

% this will contain the mean of the training set

images.data_mean = zeros(H, W, CH, 'single');

% a label per image

images.labels = zeros(1, N);

% vector indicating to which set an image belong, i.e.,
% training, validation, etc.

images.set = zeros(1, N);

numImgsTrain = 0;
% loading positive samples

for i=1:numel(positives)
  
im = single(imread([folder '/cat/', positives(i).name]));

  
images.data(:,:,:, i) = im;
  
images.labels(i) = 2;

% in this case we select the set (train/val) randomly
  

if(randi(10, 1) > 6) % 60% for training and 40% for validation

    
images.set(i) = 1;
  
  
images.data_mean = images.data_mean + im;
  
  
numImgsTrain = numImgsTrain + 1;
  

else
  
  
images.set(i) = 2;

  
end

end

% loading negative samples

for i=1:numel(negatives)

  
im = single(imread([folder '/dog/', negatives(i).name]));

  
images.data(:,:,:, NPos+NNeu+i) = im;

  
images.labels(NPos+NNeu+i) = 1;

% in this case we select the set (train/val) randomly

  
if(randi(10, 1) > 6)
  
  
images.set(NPos+NNeu+i) = 1;
  
  
images.data_mean = images.data_mean + im;

    
numImgsTrain = numImgsTrain + 1;

  
else

    
images.set(NPos+NNeu+i) = 2;

  
end

end

% loading neutral samples

for i=1:numel(neutral)

  
im = single(imread([folder '/rat/', neutral(i).name]));

  
images.data(:,:,:, NPos+NNeg+i) = im;
  

images.labels(NPos+NNeu+i) = 3;

% in this case we select the set (train/val) randomly

  
if(randi(10, 1) > 6)

    
images.set(NPos+NNeg+i) = 1;

    
images.data_mean = images.data_mean + im;

    
numImgsTrain = numImgsTrain + 1;

  
else

    
images.set(NPos+NNeu+i) = 2;

  
end

end

% let's finish to compute the mean

images.data_mean = images.data_mean ./ numImgsTrain;

% now let's add some randomization 

indices = randperm(N);

images.data(:,:,:,:) = images.data(:,:,:,indices);

images.labels(:) = images.labels(indices);

images.set(:) = images.set(indices);

imdb.meta = meta;

imdb.images = images;

end

I'm trying to give labels '1', '2', '3' for 'cat', 'dog' and 'rat' respectively. When i run the function, all the other meta-parameters including the number of sets (train, val), labels ('cat', 'dog', 'rat') are declared properly however the 'labels' and 'sets' are improperly assigned to the images. I'm pretty sure I am committing mistake in assigning the labels and sets inside the 'for' loop.

This is an issue with creating a structure and I'm sure the experts out here can solve my problem in no time. Kindly help rectifying the same with the modified peice of code. Thanks for your assistance.

[Image Analyst]

You don't need to double space your code. You can higlight and click the "{}Code" button. See http://www.mathworks.com/matlabcentral/answers/13205#answer_18099

[Sivaramakrishnan Rajaraman]

Many thanks for that suggestion. I will follow it further. Kindly help me resolve the issue with the code. Thanks once again.

[Walter Roberson]

Your code is hard to read because it is not indented. I recommend that you go into the MATLAB editor, use control A to select all of the code, and click on the smart indentation icon. Then copy that into a post here; then select the code and click on the {} button.

[Sivaramakrishnan Rajaraman]

Hi Roberson,

Here is the code after the smart indentation. Kindly help with the issue.

    function imdb = createIMDB_RAM1(folder)
    % imdb is a matlab struct with several fields, such as:
    %  - images: contains data, labels, ids dataset mean, etc.
    %  - meta: contains meta info useful for statistics and visualization
    %  - any other you want to add
    imdb = struct();

    % let's assume we have a folder with two
    % subfolders "cat" "nocat" containing images
    % for a binary problem
    positives = dir([folder '/cat/*.jpg']);
    negatives = dir([folder '/dog/*.jpg']);
    neutral = dir([folder '/rat/*.jpg']);
    imref = imread([folder '/cat/', positives(1).name]);
    [H, W, CH] = size(imref);

    % number of images
    NPos = numel(positives);
    NNeg = numel(negatives);
    NNeu = numel(neutral);
    N = NPos + NNeg + NNeu;

    % we can initialize part of the structures already
    meta.sets = {'train', 'val'};
    meta.classes = {'cat', 'dog', 'rat'};

    % images go here
    images.data = zeros(H, W, CH, N, 'single');
    % this will contain the mean of the training set
    images.data_mean = zeros(H, W, CH, 'single');
    % a label per image
    images.labels = zeros(1, N);
    % vector indicating to which set an image belong, i.e.,
    % training, validation, etc.
    images.set = zeros(1, N);

    numImgsTrain = 0;
    % loading positive samples
    for i=1:numel(positives)
        im = single(imread([folder '/cat/', positives(i).name]));
        images.data(:,:,:, i) = im;
        images.labels(i) = 2;

        % in this case we select the set (train/val) randomly
        if(randi(10, 1) > 6)
            images.set(i) = 1;
            images.data_mean = images.data_mean + im;
            numImgsTrain = numImgsTrain + 1;
        else
            images.set(i) = 2;
        end
    end

    % loading negative samples
    for i=1:numel(negatives)
        im = single(imread([folder '/dog/', negatives(i).name]));
        images.data(:,:,:, NPos+i) = im;
        images.labels(NPos+i) = 1;

        % in this case we select the set (train/val) randomly
        if(randi(10, 1) > 6)
            images.set(NPos+i) = 1;
            images.data_mean = images.data_mean + im;
            numImgsTrain = numImgsTrain + 1;
        else
            images.set(NPos+i) = 2;
        end
    end

    % loading neutral samples
    for i=1:numel(neutral)
        im = single(imread([folder '/rat/', neutral(i).name]));
        images.data(:,:,:, NPos+NNeg+i) = im;
        images.labels(NPos+NNeg+i) = 3;

        % in this case we select the set (train/val) randomly
        if(randi(10, 1) > 6)
            images.set(NPos+NNeg+i) = 1;
            images.data_mean = images.data_mean + im;
            numImgsTrain = numImgsTrain + 1;
        else
            images.set(NPos+NNeg+i) = 2;
        end
    end

    % let's finish to compute the mean
    images.data_mean = images.data_mean ./ numImgsTrain;

    % now let's add some randomization
    indices = randperm(N);
    images.data(:,:,:,:) = images.data(:,:,:,indices);
    images.labels(:) = images.labels(indices);
    images.set(:) = images.set(indices);

    imdb.meta = meta;
    imdb.images = images;

    end

[Walter Roberson]

Your current code assigns label 1 for negatives (dog), 2 for positive (cat) and 3 for neutral. Is that different from what you want?

[Sivaramakrishnan Rajaraman]

Hi Roberson, It is exactly what I want. I just want to know whether the lines:

for i=1:numel(negatives)
      im = single(imread([folder '/dog/', negatives(i).name]));
      images.data(:,:,:, NPos+i) = im;
      images.labels(NPos+i) = 1; %(for the negative examples)

and

for i=1:numel(neutral)
      im = single(imread([folder '/rat/', neutral(i).name]));
      images.data(:,:,:, NPos+NNeg+i) = im;
      images.labels(NPos+NNeg+i) = 3; %for the neutral examples

are the right way of declaring the data for the 'negative' and 'neutral' classes. In the for loop for 'negatives', do I need to consider 'NNeg' as well, in declaring images.data(:)? Similarly, in the for loop for 'neutral', do I need to do any other modifications in images.data(:)? Am I making the declaration right for the images.data (:) in both the for loops?

[Walter Roberson]

Those lines look okay.

"In the for loop for 'negatives', do I need to consider 'NNeg' as well, in declaring images.data(:)"

The current code is correct for placing all of the images together into images.data() with the positive images first, then the negative images, then the neutral ones.

[Greg Heath]

The classical approach to NN classification of non-overlapping classes is to use {0,1} unit vector targets.

Then the outputs can be interpreted as conditional posterior probabilities.

See any text on pattern recognition.

Hope this helps.

Greg

[Sivaramakrishnan Rajaraman]

Hi Roberson, Many thanks for your assistance. As you mentioned, I found that the above code is working fine to create a structure of three classes with their respective labels.