How to increase the computation time of this code?

Hi, I am working on computing some features of an image data set and saving the features for later use. Below is the code:

tic
l = 9907 % size of image data set
% pre-allocating space for variables in the for loop
Icolor = cell(1,l);
Iwave = cell(1,l);
IglrlFeatures = cell(1,l);

for i = 1:l % l = size of image data set = 9907
    IDB{1,i} = imread(strcat(path,strcat(num2str(i),'.jpg')));
    Icolor{1,i} = colorMoments(IDB{1,i}); % 6-features in each cell
    Iwave{1,i} = waveletTransform(IDB{1,i}); % 8-features in each cell
    IglrlFeatures{1,i}  = textureFeatures(IDB{1,i}); % 44-features in each cell
    ICW{1,i} = [Icolor{1,i} Iwave{1,i} IglrlFeatures{1,i}];
end
toc

Here the computation time for each function on single image is:

colorMoments(single_image) = Elapsed time is 0.009689 seconds.

waveletTransform(single_image) = Elapsed time is 0.018069 seconds.

textureFeatures(single_image) = Elapsed time is 0.022902 seconds.

l = data set size = 9907 images

Computational times for different data set sizes (l):

l = 10; Elapsed time is 0.402629 seconds.

l = 100; Elapsed time is 2.233971 seconds.

l = 1000; Elapsed time is 21.178395 seconds.

l = 2000; Elapsed time is 44.510071 seconds.

l = 5000; Elapsed time is 111.393866 seconds.

l = 9907; Elapsed time is 238.924998 seconds. approximately (~4 mins)

I want to decrease this computational time, any suggestions?

Thanks, Gopi

[Jan]

Do not use "patH" as name of a variable. This might lead to strange effects during debugging, because this is an important Matlab function.

Please use the profiler to find the bottleneck of your code. If it is imread(), optimizing the calculation will not be efficient.

Do you have the parallel processing toolbox?

[Gopichandh Danala]

Hi Jan Simon,

I tried to change the variable name "path" to something else, and surprisingly the computation time for 10,000 image dataset reduced from 238.924998 secs to 91.339700 secs.

I was surprised great information about the "path" variable I used it a lot till now will change it from now on..Thanks

Regarding parallel processing toolbox..is it parallel pool in MATLAB 2016 if so I have it, how can I use it to improve the performance more?

Next, I heard parfor is a lot better than for in some cases, but I used cells in for loop and it gives me error and have no idea how to fix that issue can you help with this issue too.

Thanks, Gopi

[Jan]

Please post the code and the error message.

[Gopichandh Danala]

Here is the final code that is working:

Icolor = cell(1,l);
Iwave = cell(1,l);
idb = cell(1,l);
tic
parfor i = 1:l
    idb{1,i} = imread(strcat(p_address,strcat(num2str(i),'.jpg')));
    Icolor{1,i} = colorMoments(idb{1,i}); % 6-features in each cell
    Iwave{1,i} = waveletTransform(idb{1,i}); % 8-features in each cell
%     IglrlFeatures{1,i}  = textureFeatures(IDB{1,i}); % 44-features in each cell
%     ICW{1,i} = [Icolor{1,i} Iwave{1,i} IglrlFeatures{1,i}];
    icw{1,i} = [Icolor{1,i} Iwave{1,i}];
end
toc

I don't see much difference between this and above code, but maybe my previous error was because of some global variables inside my GUI functions. I think they messed it up.

Performace improvement:

for loop: 92secs

parfor loop: 31secs

Really helpful, But one last question when we start a parallel pool in Matlab2016, it starts 4 workers running what does it mean and does this number reflect performance speed. like if workers are more does it mean faster or increased performance

Answer 1

[Swaroop Mishra]

Number of workers by default is equal to the number of cores in your system. You can change it inside Home-Parallel-Manage Cluster Profiles tab of MATLAB Toolstrip.

If you increase the number of workers, based on how parallel your algorithm can work, the performance can vary. There might be a bottleneck on sequential part of the algorithm. You might not be able to improve it.

If you have 4 workers, it does not mean that you will get exactly 4x speed as the sequential code. It means that you will get better computation speed for part of the code that is parallelizable.

[Gopichandh Danala]

Ok, Thanks for explaining parallel pool