Current learning AlexNet Deep Learning, I am pretty sure my image size are correct but it seems that it doesn't acknowledge it.
1 Ansicht (letzte 30 Tage)
Ältere Kommentare anzeigen
The code:
alex = alexnet;
layers = alex.Layers;
layers(23) = fullyConnectedLayer(3);
layers(25) = classificationLayer;
allImages = imageDatastore('common weed images','IncludeSubfolders',true,'LabelSource','foldernames');
[trainingImages, testImages] = splitEachLabel(allImages,0.8,'randomized');
opts = trainingOptions("sgdm",'InitialLearnRate', 0.001 ,'MaxEpochs', 30,'MiniBatchSize', 64);
myNet = trainNetwork(trainingImages, layers, opts);
The ERROR:
Error using trainNetwork
The training images are of size 277×277×3 but the input layer expects images of size 227×227×3.
Error in transferLearning (line 18)
myNet = trainNetwork(trainingImages, layers, opts);
Tried using with augmentedimagedatastore:
alex = alexnet;
layers = alex.Layers;
layers(23) = fullyConnectedLayer(3);
layers(25) = classificationLayer;
allImages = imageDatastore('common weed images','IncludeSubfolders',true,'LabelSource','foldernames');
[trainingImages, testImages] = splitEachLabel(allImages,0.8,'randomized');
trainingImages = augmentedImageDatastore([277 277],trainingImages)
testImages = augmentedImageDatastore([277 277],testImages)
opts = trainingOptions('sgdm', 'InitialLearnRate', 0.001,...
'MaxEpochs', 1, 'MiniBatchSize', 1);
myNet = trainNetwork(trainingImages, layers, opts);
Still has the same ERROR:
Error using trainNetwork
The training images are of size 277×277×3 but the input layer expects images of size 227×227×3.
Error in transferLearning (line 18)
myNet = trainNetwork(trainingImages, layers, opts);
0 Kommentare
Antworten (1)
Walter Roberson
am 14 Sep. 2022
Notice 277 compared to 227.
I suggest that you use an augmentedDatastore to implement automatic resizing to 227
3 Kommentare
Walter Roberson
am 14 Sep. 2022
An augmented store works like a filter on the image data store. You augment the image store.
Siehe auch
Kategorien
Mehr zu Image Data Workflows finden Sie in Help Center und File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!