Out of Memory, The size of the indicated variable or array appears to be changing with each loop iteration. Undefined function or variable 'Feat1'.

clear all
clc



Feat1=[];
Feat2=[];


PF = 'D:\Project\DB1\train\' % Parent Folder
SFL = dir(PF) % List the sub folders
[mp, np] = size(SFL); % compute size = number of subfolders & files & . & ..
csf=0; % counter of JUST subfolders found in PF
for i=3:mp
%% keep only folders:
k = strfind(SFL(i).name,'.');
if isempty(strfind(SFL(i).name,'.'))
csf = csf +1; % one sub folder found
SFN = SFL(i).name ;% extract his name
tifList = ls(sprintf('%s%s%s%s',PF,SFN,'\','*.tif')); % list all jpg files
[ms, ns] = size(tifList); % ms = number of image files found

%% Processing for each tif file:
for j=1:ms
featureVector=[];
Feat=[];

tifFileName = tifList(j,:); % extract name of tif file
%for currentImage1 = 1:ms
IM=imread([PF SFN '\' tifFileName]);


%hold on;

I = imresize(IM,[200,50]);

featureVector = hog_feature_vector(I);
Feat = featureVector';
Feat1 =  [Feat1 Feat];
end

PF_SFN_imgName = sprintf('%s%s%s',PF,SFN,'\',tifFileName)
%imshow(PF_SFN_imgName)
%pause(1)
end

end


  save('featurs_T','Feat1');
  
  
  PF = 'D:\Project\DB\test\'; % Parent Folder
SFL = dir(PF) ;% List the sub folders
[mp1, np] = size(SFL); % compute size = number of subfolders & files & . & ..
csf1=0; % counter of JUST subfolders found in PF
for i=3:mp1
%% keep only folders:
k = strfind(SFL(i).name,'.');
if isempty(strfind(SFL(i).name,'.'))
csf1 = csf1 +1; % one sub folder found
SFN = SFL(i).name ;% extract his name
tifList = ls(sprintf('%s%s%s%s',PF,SFN,'\','*.tif')); % list all jpg files
[ms1, ns] = size(tifList); % ms = number of image files found

%% Processing for each tif file:
for s=1:ms1
featureVector1=[]
FeatTest=[]
tifFileName = tifList(s,:) % extract name of tif file
%for currentImage1 = 1:ms
IMT =imread([PF SFN '\' tifFileName])
IT = imresize(IMT,[200,50]);

featureVector1 = hog_feature_vector(IT);
FeatTest =  featureVector1';
Feat2 = [Feat2 FeatTest];


%hold on;
end
PF_SFN_imgName = sprintf('%s%s%s',PF,SFN,'\',tifFileName)
%

%imshow(PF_SFN_imgName)
%pause(1)
end
end


save('featurs_S','Feat2');
%----------------------

   

[Roy Kadesh]

How do you prevent this from growing out of control (and taking a lot of time due to dynamic expansion)?

Feat1 =  [Feat1 Feat];

[sun rise]

Intel (R) Core(TM) i7 - 10510U CPU @ 1.80GHz 2.30GHz

RAM 12.0 GB

System type 64 bit

GPU free

Answer 1

[Jan]

You cannot get multiple errors, because Matlab stops at the first error already. So please post a full copy of the complete message. Then the readers do not have to guess, what the problem is.

A simplification of your code:

Folder   = 'D:\Project\DB1\train\';
FileList = dir(fullfile(Folder, '**', '*.tif'));
Feature  = cell(1, numel(FileList));  % Pre-allocation
for iFile = 1:numel(FileList)
   File = fullfile(FileList(iFile).folder, FileList(iFile).name);
   Img  = imread(File);
   Img  = imresize(Img, [200, 50]);
   Feature{iFile} = hog_feature_vector(Img).';  % [EDITED] Transpose added
end

% Maybe:
Feat1 = cat(1, Feature{:});  % Or cat(2, ...) ?!
save('featurs_T.mat', 'Feat1');

Parsing the output of LS is far too complicated. There is no guarantee, that "." and ".." are the 1st two outputs of the DIR command.

[sun rise]

the first error out of memory. Then I added

featureVector=[];

Feat=[];

became

Undefined function or variable 'Feat1'.

[Jan]

Inthe code you have posted, there are two lines:

Feat1=[];
Feat2=[];

Afterwards Feat1 is defined and I cannot understand, why the message appears.

Your lines

featureVector=[];
Feat=[];

Are useless, because you overwrite these variables directly.

Does my code work?

[sun rise]

Error using cat

Out of memory. Type "help memory" for your options.

Error in HOG_NEW (line 14)

Feat1 = cat(1, Feature{:}); % Or cat(2, ...) ?!

[sun rise]

clear all
clc

Folder   = 'D:\Project\DB1\train\';
FileList = dir(fullfile(Folder, '**', '*.tif'));
Feature  = cell(1, numel(FileList));  % Pre-allocation
for iFile = 1:numel(FileList)
   File = fullfile(FileList(iFile).folder, FileList(iFile).name);
   Img  = imread(File);
   Img  = imresize(Img, [400, 200]);
   Feature{iFile} = hog_feature_vector(Img);
end
% Maybe:
Feat1 = cat(1, Feature{:});  % Or cat(2, ...) ?!
save('featurs_T.mat', 'Feat1');


Folder2   = 'D:\Project\DB1\test\';
FileList2 = dir(fullfile(Folder2, '**', '*.tif'));
Feature2  = cell(1, numel(FileList2));  % Pre-allocation
for iFile2 = 1:numel(FileList2)
   File2 = fullfile(FileList2(iFile2).folder, FileList2(iFile2).name);
   Img2  = imread(File2);
   Img2  = imresize(Img2, [400, 200]);
   Feature2{iFile2} = hog_feature_vector(Img2);
end
% Maybe:
Feat2 = cat(1, Feature2{:});  % Or cat(2, ...) ?!
save('featurs_S.mat', 'Feat2');

[Jan]

Note: The "clear all" is a waste of time only. It deletes all loaded files from the memory and the reloading from the slow disk need a lot of time. Use functions to keep your workspace clean.

If cat(1, Feature2{:}) exhausts your RAM, the solution is to close other applications or better to install more RAM. How many files do you process and how large are the feature vectors?

[sun rise]

IFN/ENT dataset, 26459 files

featurs vectors are 19404

[Jan]

Then Feat2 needs 4.11 GB of RAM. This must be available in a contiguous block.

Storing such a huge data set in one block is not useful. What about storing the data in a binary file?

FID = fopen(FileName, 'w');
if FID == -1, error('Cannot open file for writing: %s', FileName); end
for k = 1:numel(Feature2)
   fwrite(FID, Feature2{k}, 'double');
end
fclose(FID)

Maybe it is a good idea to store some data on the top of the file, e.g. the dimensions of the array.

But if you want to load the data, you still need 4.11GB of free RAM.