No appropriate method, property, or field 'Files' for class 'augmentedImageDatastore'.
4 次查看(过去 30 天)
显示 更早的评论
Hi,
I am trying to apply "Train Deep Learning Network to Classify New Images". I did test exactly according to
But I always got some problem as:
"No appropriate method, property, or field 'Files' for class 'augmentedImageDatastore'.
Error in googlenetJIAGUI (line 88)
valFrequency = floor(numel(augimdsTrain.Files)/miniBatchSize);"
When I ran:
%Train Network
pixelRange = [-30 30];
scaleRange = [0.9 1.1];
imageAugmenter = imageDataAugmenter( ...
'RandXReflection',true, ...
'RandXTranslation',pixelRange, ...
'RandYTranslation',pixelRange, ...
'RandXScale',scaleRange, ...
'RandYScale',scaleRange);
augimdsTrain = augmentedImageDatastore(inputSize(1:2),imdsTrain, ...
'DataAugmentation',imageAugmenter);
augimdsValidation = augmentedImageDatastore(inputSize(1:2),imdsValidation);
%Specify the training options.
miniBatchSize = 10;
valFrequency = floor(numel(imdsValidation.Files)/miniBatchSize); %%valFrequency = floor(numel(augimdsTrain.Files)/miniBatchSize);
options = trainingOptions('sgdm', ...
'MiniBatchSize',miniBatchSize, ...
'MaxEpochs',6, ...
'InitialLearnRate',3e-4, ...
'Shuffle','every-epoch', ...
'ValidationData',augimdsValidation, ...
'ValidationFrequency',valFrequency, ...
'Verbose',false, ...
'Plots','training-progress');
I check inside of "augimdsTrain", there is no "Files". When I replace with the "valFrequency = floor(numel(augimdsTrain.Files)/miniBatchSize);" with "valFrequency = floor(numel(imdsValidation.Files)/miniBatchSize)" . "valFrequency = floor(numel(augimdsTrain.Files)/miniBatchSize);" It seems to work.
Any body could tell me whether I am right?
Best regards,
jiagui
4 个评论
Ameer Hamza
2020-7-2
On which platform are you running MATLAB R2020a? When I run it, the augmentedImageDatastore does have a field named Files.
>> disp(augimdsTrain)
augmentedImageDatastore with properties:
NumObservations: 55
Files: {55×1 cell}
AlternateFileSystemRoots: {}
MiniBatchSize: 128
DataAugmentation: [1×1 imageDataAugmenter]
ColorPreprocessing: 'none'
OutputSize: [227 227]
OutputSizeMode: 'resize'
DispatchInBackground: 0
采纳的回答
Walter Roberson
2020-7-1
Hack way:
old_warning_state = warning('off', 'MATLAB:structOnObject');
temp = struct(augimdsValidation);
warning(old_warning_state);
valFrequency = floor(numel(temp.DatastoreInternal.Files)/miniBatchSize);
3 个评论
Ameer Hamza
2020-7-2
Jiagui, try the following line
valFrequency = floor(numel(temp.DatastoreInternal.NumObservations)/miniBatchSize);
% or maybe this
valFrequency = floor(numel(temp.Files)/miniBatchSize);
更多回答(0 个)
另请参阅
类别
在 Help Center 和 File Exchange 中查找有关 Image Data Workflows 的更多信息
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!