Error while loading an object of class 'nnet.cnn.​layer.Imag​eInputLaye​r': Reference to non-existent field 'IsAverage​ImageMeanP​erChannel'​.

Hi, I have a little bit of an interesting problem, my class had a challenge to see who could make an effective image classifier. I opted to go with a neural network, and it work very well when ran on my computer. However when I turned in the folder with the neccessary components, my professor told me that it wouldn't run when they tried. She emailed me the following error:

Warning: While loading an object of class 'nnet.cnn.layer.ImageInputLayer':
Reference to non-existent field 'IsAverageImageMeanPerChannel'.
> In pet_classifier (line 2)
 In SammyMain (line 11)
Warning: Directory already exists.
> In pet_classifier (line 8)
 In SammyMain (line 11)
Dot indexing is not supported for variables of this type.

Error in nnet.internal.cnn.SeriesNetwork/setupNetworkForHostPrediction (line 511)
               this.Layers{el} = this.Layers{el}.setupForHostPrediction();

Error in SeriesNetwork/getPredictionEnvironment (line 870)
               privateNetwork = this.PrivateNetwork.setupNetworkForHostPrediction();

Error in SeriesNetwork/predict (line 312)
               predictEnvironment = this.getPredictionEnvironment(executionEnvironment);

Error in SeriesNetwork/classify (line 564)
           scores = this.predict( X, varargin{:} );

Error in pet_classifier (line 28)
   test = classify(net, testimages);

Error in SammyMain (line 11)
yguess = pet_classifier(Xtest);

The code for my program is as follows:

function yguess = pet_classifier(Xtest)
    loadednet = load('petclassifiernet.mat', 'net');
    
    net = loadednet.net;
    
    [rtest,~] = size(Xtest);

    mkdir('testimgs');
    testpath = [pwd '/testimgs'];
    for i = 1:rtest
        image = mat2gray(reshape(Xtest(i,:),64,64));
        if i < 10
            num = ['000', int2str(i)];
        elseif i < 100
            num = ['00', int2str(i)];
        elseif i < 1000
            num = ['0', int2str(i)];
        else
            num = int2str(i);
        end
        filename = sprintf('testimage%s.jpg', num);
        imgfile = fullfile(testpath, filename);
        imwrite(image,imgfile);
    end
    
    testimages = imageDatastore(testpath);
    
    test = classify(net, testimages);

    [r,~] = size(test);

    yguess = [];

    for i = 1:r
        if(test(i,1) == 'cat')
            yguess = [yguess; -1];
        elseif(test(i,1) == 'dog')
            yguess = [yguess; 1];
        else
            disp('wrong');
        end
    end

end    

And the folder I turned in contained my neural net saved as 'petclassifiernet.mat' and the function with the code as seen above. Does anyone know why the code runs on my computer but not when I turned it in?

[Maria Delgado]

Hello! Did you find a solution? I am encountering the same problem.

Thanks!

[Samuel Tomp]

I forget the exact details, but I believe the issue was that my professor was running an older version of MATLAB. I believe I was running 2019a at the time and she was running some version older than 2018, I forget which exactrly.