How to read image one by one from folder and make prediction save it in CSV format

I want to read image automatically from folder one by one and pass it to trained model one by one after some delay

I am implement the following code using GUI (push button) but it take input image from user and print the prediction

I want to make it automatically take all images from folder and processed one by one

function pushbutton1_Callback(hObject, eventdata, handles)
% hObject    handle to pushbutton1 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    structure with handles and user data (see GUIDATA)
global ImageFile filepath
[filename, filepath] = uigetfile({'*.*';'*.jpg';'*.png';'*.bmp'},'Select Image File');    
fullname = [filepath filename];
      
  % ----now read that image (fullname)
  I = imread(fullname);
  I = repmat(I,[1 1 3]);  
  % ----now display that image (fullname)
  %resizePos = get(handles.axes1,'Position');
  %ImageFile = imresize(ImageFile, [resizePos(3) resizePos(3)]); 
 
  axes(handles.axes1);
  %imagesc(ImageFile);
  imshow(I)
  
  % ----clear axes scale
  axis off
 
    net = load('11ClassesResnet.mat')
    net=net.trainednet
    inputSize = net.Layers(1).InputSize;
    I = imresize(I,inputSize(1:2));
    %-----------------------Test-------------------
    YPred_Test = classify(net,I)
    n = string(YPred_Test(1));
    set(handles.edit1,'string',n);

Answer 1

[Hiro Yoshino]

I would use "imageDatastore".

This data format is dedicated for that kind of problem.

https://www.mathworks.com/help/matlab/ref/matlab.io.datastore.imagedatastore.html?searchHighlight=imageDatastore&s_tid=srchtitle_imageDatastore_1

Use is quite straight foward: 1) point the folder that contains your images 2) read(imds) return a pointer to an image one by one

You can also extract the path information from this type of variable.

[hammad younas]

@Hiro i am using the following code it work but it read ony png extenstion i want to read all files in folder.

and process one image after 1 second

path_directory='C:\Users\ASUS\Documents\MATLAB\Examples\R2021b\phased\ModClassificationOfRadarAndCommSignalsExample\Dataset\Dataset'; % 'Folder name'
   original_files=dir([path_directory '/*.png']); 
   
  % original_files=dir( fullfile(path_directory ,['*' ext]) );
   for k=1:length(original_files)
      filename=[path_directory '/' original_files(k).name];
      I=imread(filename);
      % Image read is done
    %%Image Operation as per your work
  
    % process x
  
      
  % ----now read that image (fullname)
  %I = imread(fullname);
  I = repmat(I,[1 1 3]);  
  % ----now display that image (fullname)
  %resizePos = get(handles.axes1,'Position');
  %ImageFile = imresize(ImageFile, [resizePos(3) resizePos(3)]); 
 
  axes(handles.axes1);
  %imagesc(ImageFile);
  imshow(I)
  
  % ----clear axes scale
  axis off
 
    net = load('11ClassesResnet.mat')
    net=net.trainednet
    inputSize = net.Layers(1).InputSize;
    I = imresize(I,inputSize(1:2));
    %-----------------------Test-------------------
    YPred_Test = classify(net,I)
    n = string(YPred_Test(1));
    set(handles.edit1,'string',n);
   end

Answer 2

[Walter Roberson]

    net = load('11ClassesResnet.mat')
    net=net.trainednet
    inputSize = net.Layers(1).InputSize;
    
path_directory='C:\Users\ASUS\Documents\MATLAB\Examples\R2021b\phased\ModClassificationOfRadarAndCommSignalsExample\Dataset\Dataset'; % 'Folder name'
original_files = dir(path_directory); 
original_files([original_files.isdir]) = [];  %remove . and .. and subfolders
   
  % original_files=dir( fullfile(path_directory ,['*' ext]) );
   for k=1:length(original_files)
       thisfile = original_files(k).name;
      filename = fullfile(original_files(k).folder, thisfile);
      try
        I = imread(filename);
      catch ME
          fprintf('File "%s" could not be read as an image\n', thisfile);
          continue;
      end
      % Image read is done
    %%Image Operation as per your work
  
    % process x
  
      
  % ----now read that image (fullname)
  %I = imread(fullname);
  I = repmat(I,[1 1 3]);  
  % ----now display that image (fullname)
  %resizePos = get(handles.axes1,'Position');
  %ImageFile = imresize(ImageFile, [resizePos(3) resizePos(3)]); 
 
  axes(handles.axes1);
  %imagesc(ImageFile);
  imshow(I)
  title(thisfile)
  
  t
  % ----clear axes scale
  axis off
 

    I = imresize(I,inputSize(1:2));
    %-----------------------Test-------------------
    YPred_Test = classify(net,I)
    n = string(YPred_Test(1));
    set(handles.edit1,'string',n);
    drawnow();
   end

[hammad younas]

@Walter Roberson Thanks for you answer can you please tell how to add delay like i want to read 2nd image after 0.5 seconds How can i add it above code. and save prediction YPred_Test in CSV with Filename

[hammad younas]

@Walter Roberson Can you Help save prediction YPred_Test with Filename in CSV in above code

[Walter Roberson]

    net = load('11ClassesResnet.mat')
    net=net.trainednet
    inputSize = net.Layers(1).InputSize;
    
path_directory='C:\Users\ASUS\Documents\MATLAB\Examples\R2021b\phased\ModClassificationOfRadarAndCommSignalsExample\Dataset\Dataset'; % 'Folder name'
original_files = dir(path_directory); 
original_files([original_files.isdir]) = [];  %remove . and .. and subfolders

num_files = length(original_files);

Filename = {original_files.name}.';
Prediction = strings(num_files,1);
file_successful = false(num_files,1);

  % original_files=dir( fullfile(path_directory ,['*' ext]) );
   for k=1:length(original_files)
       thisfile = original_files(k).name;
      filename = fullfile(original_files(k).folder, thisfile);
      try
        I = imread(filename);
      catch ME
          fprintf('File "%s" could not be read as an image\n', thisfile);
          continue;
      end
      % Image read is done
    %%Image Operation as per your work
  
    % process x
  
      
  % ----now read that image (fullname)
  %I = imread(fullname);
  I = repmat(I,[1 1 3]);  
  % ----now display that image (fullname)
  %resizePos = get(handles.axes1,'Position');
  %ImageFile = imresize(ImageFile, [resizePos(3) resizePos(3)]); 
 
  axes(handles.axes1);
  %imagesc(ImageFile);
  imshow(I)
  title(thisfile)
  
  t
  % ----clear axes scale
  axis off
 

    I = imresize(I,inputSize(1:2));
    %-----------------------Test-------------------
    YPred_Test = classify(net,I);
    file_successful(k) = true;
    n = string(YPred_Test(1));
    Prediction = n;
    set(handles.edit1,'string',n);
    pause(0.5)
   end
   
   Filename = Filename(file_succesful);
   Prediction = Prediction(file_succesful);
   T = table(Filename, Prediction);
   writetable(T, 'Predictions.csv');

[hammad younas]

@Walter Roberson Getting error .

Error using string

No constructor 'string' with matching signature found.

Error in DeepLearning>pushbutton1_Callback (line 133)

Prediction = string(num_files,1);

Error in gui_mainfcn (line 95)

feval(varargin{:});

Error in DeepLearning (line 42)

gui_mainfcn(gui_State, varargin{:});

Error in matlab.graphics.internal.figfile.FigFile/read>@(hObject,eventdata)DeepLearning('pushbutton1_Callback',hObject,eventdata,guidata(hObject))

Error while evaluating UIControl Callback.

[Walter Roberson]

I corrected the code.

[hammad younas]

@Walter Roberson not working

The logical indices contain a true value outside of the array bounds.

Error in DeepLearning>pushbutton1_Callback (line 230)

Prediction = Prediction(file_successful);

Error in gui_mainfcn (line 95)

feval(varargin{:});

Error in DeepLearning (line 42)

gui_mainfcn(gui_State, varargin{:});

Error in matlab.graphics.internal.figfile.FigFile/read>@(hObject,eventdata)DeepLearning('pushbutton1_Callback',hObject,eventdata,guidata(hObject))

Error while evaluating UIControl Callback.

[hammad younas]

@Walter Roberson Its not working i have modified it too Can you please check

[Walter Roberson]

I do not seem to find a copy of 11ClassResNet.mat anywhere.

The directory name you give appears to be associated with the example https://www.mathworks.com/help/phased/ug/modulation-classification-of-radar-and-communication-waveforms-using-deep-learning.html which is an example about signals not about images so I am not clear as to what you are doing.

Ah... the line

    Prediction = n;

should be

   Prediction(k) = n;

[hammad younas]

bascially the code save the signal in image form. and trained the squeezenet model on it