showing: non existent field 'ClassNames'. Can someone please help me?

function ObjectRecognizerLive_shape(trainedClassifier,bag)

[fig, ax1, ax2] = figureSetup(trainedClassifier);

% Start webcam
wcam = webcam;

% Run live car detection
while ishandle(fig)
    % Step 1: Get Next Frame
    img = snapshot(wcam);
    grayimg = rgb2gray(img);

      % Step 2: Extract Features
    shapeFeatures = double(encode(bag,grayimg));

      % Step 3: Predict car using extracted features
    [imagepred, probabilities] = predict(trainedClassifier,shapeFeatures);

      % Step 4: Plot Results
      try
          imshow(insertText(img,[640,1],upper(cellstr(imagepred)),...
              'AnchorPoint','RightTop','FontSize',50,'BoxColor','Green',...
              'BoxOpacity',0.4),'Parent',ax1);    
          ax2.Children.YData = probabilities;
          ax2.YLim = [0 1];
      catch err
      end
      drawnow
  end 

function cname = getClassifierName(trainedClassifier)
% getClassifierName extracts name of the classifier from a trained model
cname = class(trainedClassifier);
if isa(trainedClassifier,'ClassificationECOC')
    cname = 'SVM';
end
pos = strfind(cname,'.');
if ~isempty(pos)
  cname = cname(pos(end)+1:end);
end

function [fig, ax1, ax2] = figureSetup(trainedClassifier)
% figureSetup sets up figure window for webcam feed and bar chart for
% classification probability
warning('off','images:imshow:magnificationMustBeFitForDockedFigure')
set(0,'defaultfigurewindowstyle','docked')
fig = figure('Name','Recogonizer','NumberTitle','off');
ax1 = subplot(2,1,1);
ax2 = subplot(2,1,2);
bar(ax2,zeros(1,numel(trainedClassifier.ClassNames)),'FaceColor',[0.2 0.6 0.8])
set(ax2,'XTickLabel',cellstr(trainedClassifier));
title(getClassifierName(trainedClassifier.ClassNames)), ylabel('Probability')
set(0,'defaultfigurewindowstyle','normal')

[Walter Roberson]

What are you passing in as your first argument to ObjectRecognizerLive_shape() ? What does class() show for it?

[Hermise Raju]

I am training an image set using the classification learner app. trainedClassifier is the imported data from the app and bag is the bag of features (extracting features). I want the classnames in trainedClassifier, but it is showing as a non existent field.

[Hermise Raju]

I am passing in the same trainedClassifier, bag and cname gives struct

[Adam]

Are you expecting the field to exist? What do you see when you look at the struct in the debugger? If it doesn't exist then it doesn't exist and I'm not sure what help we can give you. Did you capitalise it correctly?

[Hermise Raju]

Yeah I am expecting this field to exist

[Khurshid Aliev]

Did you solve the problem? Are you able to see ClassNames field? If so can you share your experinece?

[ing mecrat]

hola tengo el mismo error si podrian ayudarme les agradeseria gracias

[Jhon Edward Rubio Ardila]

Hola tengo el mismo error podran ayudarme.

Gracias

[Fernando Ortega Loza]

Hi, te problem is the Classification Learner, the rsesult of this is a struct, into the object we have the Trained Classifier, into we have de ClassNames

[Image Analyst]

@Fernando Ortega Loza When you export the trained model it goes into the workspace with the name you give it, by default "trainedModel'. When you call save() to save it into a mat file,

save('My Classifier.mat', 'trainedModel');

and then recall the mat file

s = load('My Classifier.mat');

s will be a structure. You can then get the trained model from the structure like this:

trainedModel = s.trainedModel

If you look at trained model, I believe one of the fields say how to use it. So then you can pass trainedModel into predict

Answer 1

[Image Analyst]

Use the debugger. It's the way the rest of us solve problems like this, and you should too. So look at where you call ObjectRecognizerLive_shape(trainedClassifier,bag) and see what you pass in for trainedClassifier to see if that class has any properties called ClassNames. It won't. It will have a field according to what classifier you used. So if you didn't save or remember the classifier you used you'll have to check the fields with fieldnames() and strcmp() or ismember() to see which class you used: ClassificationSVM, CompactClassificationSVM, ClassificationECOC, or CompactClassificationECOC.

Here is what the help says:

Find the name of the classification model object in the exported struct. Examine the fields of the struct to find the model name, for example, C.ClassificationSVM, where C is the name of your struct, e.g., trainedClassifier. Model name depends on what type of SVM you trained (binary or multiclass) and whether you exported a compact model or not. Models can be ClassificationSVM, CompactClassificationSVM, ClassificationECOC, or CompactClassificationECOC.

Like I said, you can use fieldnames() and strfind() for that.

[Hermise Raju]

Thank you so much

[Hermise Raju]

Also can you please tell me what goes inside cellstr? It needs a character vector

[Walter Roberson]

is effectively:

function result = cellstr(c)
  numrow = size(c,1);
  result = cell(numrow, 1);
  for row = 1 : numrow
    result{row} = deblank(c(row,:));
  end

That is, you pass in a character array with one or more rows, and it turns it into a cell array of strings with trailing blanks removed.

The opposite of cellstr() is char() applied to a cell array of strings, which creates a character array with one or more rows, with all rows the length of the widest string from the cell, with the shorter strings blank padded on the right.

[Hermise Raju]

Thank you so much

[Hermise Raju]

[imagepred, probabilities] = predict(trainedClassifier,shapeFeatures)

when I run this, it is showing that no valid system or data is specified. My system is the trainedClassifier from the classification learner app, and my data is the shapeFeature from features extracted from bag of features. Can you please help me why it is showing that error?