How to take image from multi folders?

Hi everybody

I have a few thousand folders, each folder contains 2 images with information in their name, i want to take randomly 1 image from each folder and put all the taken images into one new folder in desktop for further calculation, please help, thank you very much.

[Rik]

Use the ** expansion for dir, after which you can use rand. What did you try?

[D. Frank]

I got the info by using dir, and still trying to find the way to pick image from each files, should i write by function ? please help thank you

[Rik]

You mean you want to read an image? If you know the file names you can just use imread.

[D. Frank]

Read image is the next step of computing these files, the main thing i want to do is taking 1 random image from each subfolder, and there are 2900 subfolders in 1 folder to add to path in matlab :((((((

Answer 1

[Monalisa Pal]

Hi,

'fetching a random image' part of the following code may help you

clearvars -global
clearvars
close all
clc

% creating destination folder
destination = 'destinationfolder';
mkdir(destination);
n_folders = 10; % say there are 10 folders
for i = 1:n_folders
    % fetching the contents of the folders
    foldername = ['folder_', num2str(i), '/'];
    path = [foldername, '*.png']; % or other imgfile extensions
    contents = dir(path); 
    
    % fetching a random image
    if numel(contents) < 2
        error([foldername, ' has only ', num2str(numel(contents)), ' image files']);
    end
    img_index = randi([1, numel(contents)], 1);
    disp(['From ', foldername, ' copying ', contents(img_index).name]);
    source = [foldername, contents(img_index).name];
    copyfile(source, destination);
end

[D. Frank]

I got the error using rand

It showed "Error using randi

First input must be a positive scalar integer value IMAX, or two integer values [IMIN IMAX] with IMIN less than or equal to IMAX."

[Walter Roberson]

One of your folders has no .png files in it. You should adjust the .png in the code to the extension that is appropriate to your situation.

If some of your directories have no files of appropriate file extension, then how would you like to select randomly from the zero files in the directory?

[D. Frank]

Thank you Mr. Walter

i put the expansion from the one below but still got the same error.

path = [foldername, '\.(jpg|gif|png|jpeg|bmp)'];

most of the subfolders contains .jpg images, thank you for ur help

[Walter Roberson]

exts = {'*.jpg', '*.gif', '*.png', '*.jpeg', '*.bmp'};

for i = 1:n_folders
    % fetching the contents of the folders
    foldername = ['folder_', num2str(i), '/'];
    contents = celllfun(@(E) dir( fullfile(foldername, E) ), exts, 'uniform', 0);
    
    if isempty(contents)
        fprintf('caution: folder "%s" had no useable images\n', foldername);
        continue;
    end
    % fetching a random image
    img_index = randi([1, numel(contents)], 1);
    filename = fullfile(foldername, contents(img_index).name);
    copyfile(filename, destination);
end

[Monalisa Pal]

Thank you so much for pointing it out. I edited the above code to provide an error check.

Could you please check the following two things?

1) What is the extension for your image files? I used .png in the path. If it is different for you, kindy edit that line.

2) Does all your folders have atleast 2 images? I understand it is difficult to check manually. I have added an error check, it will tell you the name of the folder which has less than 2 images.

Let me know if the fix works or if it gives more errors.

[D. Frank]

Thank you, Ms. Monalisa Pal and Mr. Walter Roberson, i belive that this one should be right

clear;
clearvars -global
clearvars
close all
clc
% creating destination folder
destination = 'Images_place';
mkdir(destination);
n_folders = 2950; % say there are 10 folders
exts = {'*.jpg', '*.gif', '*.png', '*.jpeg', '*.bmp'};
for i = 1:n_folders
    % fetching the contents of the folders
    foldername = ['Bigfolder', num2str(i), '/'];
    contents = cellfun(@(E) dir( fullfile(foldername, E) ), exts, 'uniform', 0);
    
    if isempty(contents)
        fprintf('caution: folder "%s" had no useable images\n', foldername);
        continue;
    end
    % fetching a random image
    img_index = randi([1, numel(contents)], 1);
    filename = fullfile(foldername, contents(img_index).name);
    copyfile(filename, destination);
end

But in the line

filename = fullfile(foldername, contents(img_index).name);

i got the error showed 'Dot indexing is not supported for variables of this type'

in the 'Bigfolder' i have 2950 subfolders and each subfolder has 2 images in the range from .jpg, .png and etc

[D. Frank]

The variables may be to large to be processed

[Walter Roberson]

    contents_parts = cellfun(@(E) dir( fullfile(foldername, E) ), exts, 'uniform', 0);
    contents = vertcat(contents_parts{:});

[Monalisa Pal]

How about this?

clearvars -global
clearvars
close all
clc

% creating destination folder
destination = 'Images_place';
mkdir(destination);
n_folders = 2950; % say there are 2950 folders
for i = 1:n_folders
    % fetching the contents of the folders
    foldername = ['Bigfolder', num2str(i), '/'];
    contents = [dir(fullfile(foldername,'*.png')); dir(fullfile(foldername,'*.jpg')); dir(fullfile(foldername,'*.gif')); dir(fullfile(foldername,'*.jpeg')); dir(fullfile(foldername,'*.bmp'))]; 
    
    % fetching a random image
    img_index = randi([1, numel(contents)], 1);
    filename = fullfile(foldername, contents(img_index).name);
    copyfile(filename, destination);
end

[D. Frank]

Dear Ms. Monalisa, the code still got the error "Error using randi

First input must be a positive scalar integer value IMAX, or two integer values [IMIN IMAX] with IMIN less than or equal to IMAX." in the line:

img_index = randi([1, numel(contents)], 1);

[Walter Roberson]

    contents = [dir(fullfile(foldername,'*.png')); dir(fullfile(foldername,'*.jpg')); dir(fullfile(foldername,'*.gif')); dir(fullfile(foldername,'*.jpeg')); dir(fullfile(foldername,'*.bmp'))]; 
    

is generally viable, but it gets to be a pain to extend as you add more image file types.

But no matter what you do, you should check that you got at least one image before trying to randomize over the number of images.

[D. Frank]

Dear Mr. Roberson, the code ran but i only got this sentence from the output section:

caution: folder "Bigfolder1/" had no useable images

caution: folder "Bigfolder2/" had no useable images

caution: folder "Bigfolder3/" had no useable images

caution: folder "Bigfolder4/" had no useable images

....

[Walter Roberson]

What do you see if you

ls Bigfolder1

Note that the code works silently, not outputing anything as it copies images, only warning when it cannot copy. You should look in destinationfolder to see whether images got copied there.

[D. Frank]

take random from each subfolders is really a pain :(((, but it is sure that most are .jpg in each subfolder

[D. Frank]

when i ran

ls Bigfolder

it shows all the different names of the subfolders such as:

.       19fv20  1g4kox  1jbndn  1mlwl4  1oxx2w  1tpnv5  209468  262s2b ...
    

and in the destination file, it is blank

[Walter Roberson]

ls Bigfolder1

and tell us what the output is ?

[Monalisa Pal]

Dear Mr. Frank,

I was not familiar with cellfun. Many thanks to Mr. Roberson for mentioning it. Since you mentioned that there are definitely 2 images in each folder, I had removed the check. But it definitely helps in debugging. Incorporating his suggestions here's how the code looks like:

clearvars -global
clearvars
close all
clc

% creating destination folder
destination = 'Images_place';
mkdir(destination);
n_folders = 2950; % say there are 2950 folders
exts = {'*.jpg', '*.gif', '*.png', '*.jpeg', '*.bmp'};
for i = 1:n_folders
    % fetching the contents of the folders
    foldername = ['Bigfolder', num2str(i), '/'];
    contents_parts = cellfun(@(E) dir( fullfile(foldername, E) ), exts, 'uniform', 0);
    contents = vertcat(contents_parts{:});    
    
    % fetching a random image
    if isempty(contents)
        fprintf(1, 'caution: folder "%s" had no useable images\n', foldername);
        continue;
    end
    img_index = randi([1, numel(contents)], 1);
    disp(['From ', foldername, ' copying ', contents(img_index).name]);
    filename = fullfile(foldername, contents(img_index).name);
    copyfile(filename, destination);
end

If this doesn't work, could you please share a screenshot of your command window and workspace? I might help in debugging the code futher.

[D. Frank]

i think that i need to add to path all the subfolders in the bigfolder one

[Monalisa Pal]

Mr. Frank, I misunderstood the directory structure earlier. Lets see if this works.

clearvars -global
clearvars
close all
clc

% creating destination folder
destination = 'Images_place';
mkdir(destination);
folder = 'Bigfolder';
sub_folders = dir(folder);
%n_folders = 2950; % say there are 2950 folders
% n_folders = numel(sub_folders) - 2;
exts = {'*.jpg', '*.gif', '*.png', '*.jpeg', '*.bmp'};
for i = 3:numel(sub_folders)
    % fetching the contents of the folders
    foldername = ['Bigfolder/', sub_folders(i).name, '/'];
    contents_parts = cellfun(@(E) dir( fullfile(foldername, E) ), exts, 'uniform', 0);
    contents = vertcat(contents_parts{:});    
    
    % fetching a random image
    if isempty(contents)
        fprintf(1, 'caution: folder "%s" had no useable images\n', foldername);
        continue;
    end
    img_index = randi([1, numel(contents)], 1);
    disp(['From ', foldername, ' copying ', contents(img_index).name]);
    filename = fullfile(foldername, contents(img_index).name);
    copyfile(filename, destination);
end

[D. Frank]

Dear Ms. Monalisa Pal and Mr. Roberson, the code ran smoothly, thank you very much for your patience and your help.