new data as a input about digit-recognition HOT 14 CLOSED

 % Error Phase
    min_error = 1:10;
    for i = 1:10
        v = targets(i,:);
        errora = v - outo;
        errorf = errora.^2;
        errorf = 0.5 * errorf;
        error = sum(errorf);
        min_error(i) = error;
    end
    [~,k] = min(min_error);
    k = k-1;
    %disp(min_error)

clear all  
close all
clc

% %%   Begin training -  NO NEED SINCE THERE IS PREPROCESSED DATA ..
% 
% % Calling Initilization.m Script
% initilization
% 
% % Calling Training.m Script
% for epoc = 1:200
%     
%     msg=['new epoch number : ',num2str(epoc)];
%     msg2=['training continues...'];
%     disp(msg)
%        disp(msg2)
%     alpha = 1/(epoc^0.5);
%     %alpha = 0.5;
%     training
% 	
% 	
% end
% 
% Accuracy
% 
% save 'C:\Users\acer\Desktop\Digit Recognition\data_epoc200.mat'
%%   Ready data load

nn = load('data_epoc200.mat');
%neural_net = nn.net;

% 
% Picture_set_1=imread(char(join( (test_folder)+"\"+files_1(4).name ) ) );
% Picture_set_2=imread(char(join( (test_folder)+"\"+files_1(6).name ) ) );
% 
% s = row_and_col_values_of_img(Picture_set_1);
% 
% s2 = row_and_col_values_of_img(Picture_set_2);

% t1=binary_convertor(s2) ;



%% lOAD External TEST PIC HERE 


test_folder = 'C:\Users\enesk\Desktop\w-s\nn-matlab\50piksel\28piksel'; % test image folder on my computer... you need to change this to where your image is..
files_1 = dir(fullfile(test_folder));   % list files in that directory...

Picture_set_1=imread(char(join( (test_folder)+"\"+files_1(4).name ) ) );    % open files_1 in matlab workspace and find out your image's order
                                                                            %in directory.because directory  has items like "." and ".." so it starts from 3.  Check it out
imshow(Picture_set_1)  %show the image  
  
t = binary_convertor(Picture_set_1)   %use it if it is a non binary image . i have tried with non binary and binary images. same result. 

s = row_and_col_values_of_img(t);  %this is my function to create  1x784 vector of image piksels. this is how mnist data was created. it says row-wise in this official site...


% get what you need to use prediction function .
weight1= nn.weight1;
weight2=nn.weight2;
target=nn.target;
bias1= nn.bias1;
bias2=nn.bias2;

 %% prediction
    
     %s=~s ;
    k = prediction(s, weight1, weight2, bias1, bias2, target)

%This is my function  for calculating 1x724  vector of image  

function [s] = row_and_col_values_of_img(test_picture)

   s=zeros(1,784) ;
    I =(test_picture)  ;
          for j=1:28  %
                for i=1:28  %
             s(1, i+ 28*(j-1) )=I(j,i); 
                 end

          end
            
          
end

% 
% quick check to see if it is truly calculates ...
%     
% at first  j =1   
%     
%                 for i=1:28  %
%              s(1, i )=I(j,i); 
%                  end
% so it just copy first row to s   ...
%     
% then   j =2
%                  for i=1:28  %
%              s(1, i+ 28) )=I(2,i); 
%                  end
%                  
%     so it copies second row to  s ; starting from s(1, 29)  until  s(1, 56)         
%     
%     
%     then   j =28
%     
%      for i=1:28  %
%              s(1, i+ 28*(27) )=I(28,i); 
%                  end
%                  
%   it copies 28th row to s  starting from s(1, (28*27) -1)  until  s(1, 28*28)...