Invalid training data. The output size (5) of the last layer doesn't match the number of classes (5). How to match output size??

1 次查看（过去 30 天）
nima aalizade
nima aalizade 2017-12-16
评论： Mihai Mihaela 2019-12-8
net=vgg16();
imds = imageDatastore(fullfile('E:\','data','labels'),...
'IncludeSubfolders',true,'FileExtensions','.dcm','LabelSource','foldernames');
labelCount = countEachLabel(imds);
trainingNumFiles = 105;
rng(1) % For reproducibility
[trainData,testData] = splitEachLabel(imds,...
trainingNumFiles,'randomize');
imageSize = [512 512 1];
numClasses = 5;
encoderDepth = 9;
lgraph = segnetLayers(imageSize,numClasses,encoderDepth);
plot(lgraph)
options = trainingOptions('sgdm','InitialLearnRate',1e-3, ...
'MaxEpochs',50,'VerboseFrequency',10);
seg = trainNetwork(imds,lgraph,options)

请先登录，再回答此问题。

采纳的回答

nima aalizade
nima aalizade 2018-2-16
编辑：nima aalizade 2018-2-16
hello,
for using SegNet, you most have pixel labeled data with image labeler. you can use this and this example to understand better.

更多回答（1 个）

abdulkader helwan
abdulkader helwan 2017-12-25
Hello.. i am having the same problem here. could u please tell me how u solved it if u did so. thanks
  4 个评论
显示 2更早的评论
nima aalizade
nima aalizade 2018-2-16
编辑：nima aalizade 2018-2-16
hello
for using SegNet, you most have pixel labeled data with image labeler. you can use this and this example to understand better.

请先登录，再进行评论。

类别

在 Help Center 和 File Exchange 中查找有关 Semantic Segmentation 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by