1000 color jpg images classification with labels

I have 1000 color jpg images called train_image_#.jpg where # = 1:1000

I want to load them then with labels '1' for my training set (which consists of only one class: labelled 1)

can anyone help me with this. thanks!

Answer 1

[Image Analyst]

See the FAQ for code examples: http://matlab.wikia.com/wiki/FAQ#How_can_I_process_a_sequence_of_files.3F

[Jerone]

I've already built the file names in sequence. What I want to do is label each of the 1000 images with a label..

i.e. if i create a label = ones([1 1000]) then how do i assign this to the 1000 rgb images of size 50x50

[Image Analyst]

I don't know what that means. You don't mean label as in connected components labeling with bwlabel() or bwconncomp(). I'm not sure why you need a label like you refer to it. The label is just the loop iterator, but why do you even need it? What are you going to do with these "labels" once you have them? And why is the array you made up no good?

[Sabanam]

@ Image Analyst

Actually he was trying to say that he wanted to make training array where 1000 images are labeled as class 1 other will fall in different class label.here, class label is not for binarization but for assigning class name.

Hope It will be the right wt i have interpret?

[Jerone]

yes, that's what i'd like!

[Jerone]

i.e. i want to perform outlier detection. train on the normal class labelled 1, and then anything that does not fit in is labelled an outlier, i.e. 0

simplest method would be something like nearest neighbours

[Image Analyst]

If I understood what Sabanam said (and Jerone agreed to), it sounds like you could "label" each file by either using movefile() to rename it with a new name where the "label number" is appended to the original file name, or just put the training and testing set images in two different folders.

However Jerone's last comment seems to indicate that some kind of image analysis would be done on the image and a number would be returned. Then you'd just assign those to a results array to keep log of what class the image was determined to be in;

myFolder = 'C:/My Images';
if ~isdir(myFolder)
  errorMessage = sprintf('Error: The following folder does not exist:\n%s', myFolder);
  uiwait(warndlg(errorMessage));
  return;
end
filePattern = fullfile(myFolder, '*.jpg');
jpegFiles = dir(filePattern);
for k = 1:length(jpegFiles)
  baseFileName = jpegFiles(k).name;
  fullFileName = fullfile(myFolder, baseFileName);
  fprintf(1, 'Now reading %s\n', fullFileName);
  imageArray = imread(fullFileName);
  % Now determine what class this image is and store the result.
  % AnalyzeSingleImage() is some function that Jerone writes.
  imageClasses(k) = AnalyzeSingleImage(imageArray);
end
% Now imageClasses is a record of what the classes are for all the images.

[Jerone]

(Note: I haven't tried what you just suggested, but I will do a bit later.)

However, just to further clarify using a similar example- See this Kaggle competition http://www.kaggle.com/c/cifar-10 (CIFAR-10 - Object Recognition in Images):

There are thousands of images in the training set, and you are also given a .csv file with two columns: the first column is the name of the image, and the second column is the class that the image belongs to.

So, if i created a .csv file with the name of my images (first column) and the class of the images (second column), how do I go about loading this in matlab?

[Image Analyst]

You can use readtable() (if you have R2014a version). You can't use csvread() - it won't work since both columns aren't numbers, so you should use readtable() which can handle columns of different types. If you have an old version of MATLAB, use textscan() or fgetl() along with sscanf().