Help in next error: Undefined variable "net" or class "net.Layers". instruction: montage(ma​t2gray(gat​her(net.La​yers(2).We​ights)));

Hello Greetings to all, I'm new to MatLab, I'm trying to emulate a CNN code but I have an error, could someone help me?

My development environment is: Win 10 in 64 bits MatLab 2017b

if true
  % start
   clc;clear;

     %%Arquitectura de la red CNN sin dropout
      tamImg=227;
      layers2 = [ ...
                % data augmentation para conseguir mas imagenes
                imageInputLayer([tamImg tamImg 3],'DataAugmentation','randcrop');       
                convolution2dLayer(11,96,'Padding',0,'Stride',4);
                reluLayer();
                crossChannelNormalizationLayer(5);
                maxPooling2dLayer(3,'Stride',2);    % filtros de 3x3 y saltos(stride) de 2 en 2
                convolution2dLayer(5,256,'Padding',2,'Stride',1);
                reluLayer();        % Rectificador Lineal --> Genera valores 0 y 1
                crossChannelNormalizationLayer(5);
                maxPooling2dLayer(3,'Stride',2);
                convolution2dLayer(3,384,'Padding',1,'Stride',1);
                reluLayer();
                convolution2dLayer(3,384,'Padding',1,'Stride',1);
                reluLayer();
                convolution2dLayer(3,256,'Padding',1,'Stride',1);
                reluLayer();
                maxPooling2dLayer(3,'Stride',2);
                fullyConnectedLayer(4096);
                reluLayer();
                fullyConnectedLayer(4096);
                reluLayer();
                fullyConnectedLayer(10); 
                softmaxLayer();
                classificationLayer()];

              %%Opciones del solver
                 opts = trainingOptions('sgdm', ...
                                       'Momentum', 0.9, ...
                                       'InitialLearnRate', 0.001, ...
                                       'LearnRateSchedule', 'piecewise', ...
                                       'LearnRateDropFactor', 0.1, ...
                                       'LearnRateDropPeriod',5, ...
                                       'L2Regularization', 0.004, ...
                                       'MaxEpochs', 50, ... 
                                       'MiniBatchSize', 16, ...
                                       'Verbose', true);

                        %%Preparar imagenes de entrenamiento en conjuntos con el mismo numero
                       CarpetaRaiz = 'c:\imgs\';
                       categorias = {'Arterio_arterial','Arterio_venosa','Borde_placentario',...
                                     'Capilar','Insercion_donante', 'Insercion_receptor',...
                                     'Membrana_interfetal','Normal','Veno_arterial','Veno_venosa'};

                       imds = imageDatastore(fullfile(CarpetaRaiz, categorias),...
                                                      'LabelSource', 'foldernames');

                         % contamos imagenes
                          numImagenes = countEachLabel(imds);
                          minimo = min(numImagenes{:,2});

                         % usamos el metodo splitEachLabel para recortar el conjunto.
                          % imds = splitEachLabel(imds, minimo, 'randomize');

                          % Contamos despues de recortar
                          % countEachLabel(imds)

                         %%Entrenar la CNN
                          tic; [net4clas23, info4clas23] = trainNetwork(imds, layers2, opts); toc;

                          %%Visualizar los filtros aprendidos en la primera capa de convolución
                              figure;
                           % gather transfiere desde GPU hacia Workspace
                         montage(mat2gray(gather(net.Layers(2).Weights)));
                          title('Pesos primera capa');

end

The error that it sends is the following: montage(mat2gray(gather(net.Layers(2).Weights)));

Error: " Undefined variable "net" or class "net.Layers".

Error in CNN_AlexNet_Fetoscopy (line 107) montage(mat2gray(gather(net.Layers(2).Weights))); "

thks in advanced for help

Best regarts

Angel lerma

Answer 1

[Joss Knight]

Well, you've called your network net4clas23, not net, so no variable of that name exists.