How can I speed up the svdd one class classifier training in prtools

My Query: Is it okay for the svdd function in the dd_tools toolbox to take around 0.15 secs to train a 95X3 dataset with one class ? I am using the default values for the parameters.

Some Context: I am basically trying to train a classifier for every pixel, based on the RGB values for that pixel in the past few (95 in this particular case) frames. For this purpose I am using the svdd classifier as outlined in this paper. The paper reports to have done the entire training in 16 seconds and hence I think I doing something terrible wrong

A request: My code is currently taking a couple of hours for the entire training and I think there is something fundamentally wrong in the way I am using the toolbox. I have read through the documentation and tutorials multiple times, but I am not being able to find out what is going wrong here. Could you please take a look and help me understand the correct usage of the functions in the toolbox.

Appended is a copy of my entire code.

The driver.m file is where I am using svdd function.

Thank you.

******************************Driver.m******************************************************

%Driver File %Taking Initial SnapShot %ADD MESSAGE FOR MOVING OUT OF THE SCREEN %BackgroundFrames is a set of 100 frames taken from the video feed of the webcamera. BackgroundFrames = BackgroundRead();

%TRAINING PHASE

SVDxy{24,32}=prmapping(); %set of SVD for pixel x,y %Iterating through ever pixel for x = 1:240 for y = 1:320 trainingVector = zeros(10,3); for i=1:10 trainingVector(i,:) = BackgroundFrames(x,y,:,i); end W_Class = target_class(trainingVector,'1'); t = cputime; SVDxy{x,y} = svdd(W_Class,0.1,5); e = cputime-t; e end

end

%TESTING PHASE

% for x=1:240 % for y= 1:320 % testingVector = zeros(5,3); % labels_pred = zeros(240,320,5); % for j=91:95, % testingVector(j-95,:) = BackgroundFrames(x,y,:,j); % end % T_Class = target_class(testingVector,'1'); % finalAnswer = T_Class*SVDxy{x,y}; % % for k =1:5 % labels_pred(x,y,k) = sum(double(finalAnswer.labels(k,:)))-646; % end % end % end

*******************BackgroundFrames.m*****************************************************

%function to read and capture the background %This function creates 10 images of the background function [BackgroundFrames] = BackgroundRead() if(~isdeployed) cd(fileparts(which(mfilename))); end clc; % Clear command window. clear; % Delete all variables. close all; % Close all figure windows except those created by imtool; imtool close all; % Close all figure windows created by imtool. workspace; % Make sure the workspace panel is showing.

 vid = videoinput('winvideo', 1, 'YUY2_320x240'); 
 set(vid, 'ReturnedColorSpace', 'RGB');
 src = getselectedsource(vid);
 src.FrameRate = '20.0000';
 vid.FramesPerTrigger = 20*5;
 start(vid);
 wait(vid,65);
 data = getdata(vid);
 BackgroundFrames = data(:,:,:,1:100);
 img = BackgroundFrames(:,:,:,100);
 imshow(img);
end