Training a network using valudationData of pixelwise labeling

Hi, I am trying to train SegNet, based on this guide: semantic segmentation , while monitoring the loss function etc. I would like to use the validationData option to see how fell the model fairs during the training. I am trying to construct my validation data according to the documentation, however it does not appear to accept it. Is it possible to use this kind of validation data?

This is how i create the validationData

 % Construct validationData from image datastore.
 validIm = readall(imdsValid);
 validLabel = readall(pxdsValid);
 validationData = cell(1,2);
 validationData{1} = cat(4, validIm{:});
 validationData{2} = cat(4, validLabel{:});

 options = trainingOptions('sgdm', ...
    'Momentum', 0.9, ...
    'InitialLearnRate', 1e-3, ...
    'L2Regularization', 0.0005, ...
    'LearnRateSchedule','piecewise',...
    'LearnRateDropFactor',0.2,...
    'LearnRateDropPeriod',5,...
    'MaxEpochs', 200, ...
    'MiniBatchSize', 128, ...
    'Shuffle', 'every-epoch', ...
    'VerboseFrequency', 1, ...
    'Plots', 'training-progress', ...
    'ValidationData', validationData, ...
    'ValidationFrequency', 100);

% Specify the class weights using a |pixelClassificationLayer|.
pxLayer = pixelClassificationLayer('Name','labels','ClassNames', tbl.Name, 'ClassWeights', classWeights)

% Update the SegNet network with the new |pixelClassificationLayer| by
% removing the current |pixelClassificationLayer| and adding the new layer.
% The current |pixelClassificationLayer| is named 'pixelLabels'. Remove it 
% using |removeLayers|, add the new one using|addLayers|, and connect the 
% new layer to the rest of the network using |connectLayers|.
lgraph = removeLayers(lgraph, 'pixelLabels');
lgraph = addLayers(lgraph, pxLayer);
lgraph = connectLayers(lgraph, 'softmax' ,'labels');
% Data augmentation
augmenter = imageDataAugmenter('RandXTranslation', [-10 10], ...
  'RandYTranslation',[-10 10]);
datasource = pixelLabelImageSource(imdsTrain,pxdsTrain, ...
 'DataAugmentation',augmenter);

[net, info] = trainNetwork(datasource, lgraph, options);

[Chris Perkins]

Hi Mattias,

What exactly do you mean by 'does not appear to accept it'? Does "trainingOptions" return an error message (if so, what error) or does "trainNetwork" not perform as you expect?

In addition, can you provide your 'validationData' cell array? You should be able to attach it as a MEX file (If the file is too large, you could remove most of the data and post it with less data, just to show the full structure).

[Salma Ali]

Hi Chris,

Kindly, I am facing the same problem with the validation data. I have done the 'checkpointpath', but I do not know how to build the 'OutputFcn' and how to pass back the results of the validated data to the training to optimize the weights?

It will be helpful to give an example of using the 'OutputFcn' with the Semantic Segmentation? As you have not show such example on your site.

https://uk.mathworks.com/help/vision/examples/semantic-segmentation-using-deep-learning.html

Regards,

[Sahar Zafari]

Hi,

I am having same problem. Could you please make an example how to define validation data in trainingOption for semantic segmentation?.

Here is the error: "Invalid validation data. Y must be a vector of categorical responses."

Thanks

[Steve Scott]

Yes, I am having this problem too, Sahar.

Is the "vector of categorical responses" per-pixel or per-image? Can you get the "vector of categorical responses" from the pixel datastore?

Answer 1

[Tony Mackenzie]

This was not working for me on 2017b but I just downloaded 2018a and tested it out with my test data using:

enddatasourcetrain = pixelLabelImageSource( imdsTrain, pxdsTrain);
datasourcetest = pixelLabelImageSource( imdsTest, pxdsTest);

 options = trainingOptions('sgdm', ...
    'Momentum', .9, ...
    'InitialLearnRate', 1e-2, ...
    'MaxEpochs', 100, ...
    'MiniBatchSize',8, ...
    'Shuffle', 'every-epoch', ...
    'VerboseFrequency', 2,...
    'Plots','training-progress',...
    'LearnRateDropFactor',0.1,...
    'ValidationData', datasourcetest, ...
    'LearnRateDropPeriod',100);

Now it is working.

[nasaruddin shaikh]

In example of Semantic Segmentation Using FCN-AlexNet on link https://www.mathworks.com/examples/matlab/community/32056-fcnalexnetexample#1 they have written that training on single GPU is done .I have COU then how to do Single GPU training ? what addition settings and softwares needed? what extra commands needed? please help me? I am working on project named automatic shadow detection using CNN .then how to do it with semantic segmenetation ? please reply

[dinthisrang daimary]

hi. somebody please help me i am facing problem showing that

Error using helperDeeplabv3PlusResnet18

Too many input arguments.

Error in resnnet (line 57)

lgraph = helperDeeplabv3PlusResnet18(imageSize, numClasses);

link i have attached here

https://in.mathworks.com/help/vision/examples/semantic-segmentation-using-deep-learning.html

[RIDVAN YAYLA]

Hello I also have a problem with the same function and the same page.

There is no any code about helperDeeplabv3PlusResnet18.m file but it is said that it is attached at the bottom of the page as a supported file.

Where is the helperDeepLabv3PlusResnet18 m file???

Do you have any idea?

[todd chelsea]

there is a same question?

Where is the helperDeepLabv3PlusResnet18 m file???

anybody could do me a favor and give the helperDeepLabv3PlusResnet18.m file.

[LJ Zhou]

Actually, this file has been already in your disk, you may try this path :

C:\Users\Administrator\Documents\MATLAB\Examples\R2019a\deeplearning_shared\SemanticSegmentationUsingDeepLearningExample