The training images are of size 2062x3545x3 but the input layer expects images of size 32x32x3.

Question

Mohamed Wassim BABA 2019-1-11

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/439377-the-training-images-are-of-size-2062x3545x3-but-the-input-layer-expects-images-of-size-32x32x3

评论： Image Analyst 2020-7-19

Hello,

I want to train a D.L network with images with "2062x3545x3" of size (they are panchromatic). I follow the steps described in the "mathworks" as described bellow but I get this error: " The training images are of size 2062x3545x3 but the input layer expects images of size 32x32x3." So what I should do in order to train my network ? Many thanks.

numFilters = 64;
filterSize = 3;
numClasses = 2;
layers = [
    imageInputLayer([32 32 3])
    convolution2dLayer(filterSize,numFilters,'Padding',1)
    reluLayer()
    maxPooling2dLayer(2,'Stride',2)
    convolution2dLayer(filterSize,numFilters,'Padding',1)
    reluLayer()
    transposedConv2dLayer(4,numFilters,'Stride',2,'Cropping',1);
    convolution2dLayer(1,numClasses);
    softmaxLayer()
    pixelClassificationLayer()
    ]
opts = trainingOptions('sgdm', ...
    'InitialLearnRate',1e-3, ...
    'MaxEpochs',100, ...
    'MiniBatchSize',64);
net = trainNetwork(trainingData,layers,opts);

3 个评论
显示 1更早的评论隐藏 1更早的评论

arun anoop m 2020-7-19

I changed all images to 32x32 by the help of

picresize.com/en/batch

but still same error

Unexpected image size: All images must have the same size.

How can i solve?

Image Analyst 2020-7-19

Try my code below. Evidently not all of your images were of that size, or else you're trying to process a file that is not an image.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Image Analyst 2020-7-19

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/439377-the-training-images-are-of-size-2062x3545x3-but-the-input-layer-expects-images-of-size-32x32x3#answer_467916

在 MATLAB Online 中打开

Below is a script I use to resize my images to 224 by 224 for Alex net. Adapt as needed.

% Resizes images to a size of 224x224, which AlexNet needs, and saves the resized images in a "Resized to 224x224" subfolder of the images folder.
% Image Analyst, March 21, 2020.
clc;    % Clear the command window.
fprintf('Beginning to run %s.m.\n', mfilename);
close all;  % Close all figures (except those of imtool.)
clear;  % Erase all existing variables. Or clearvars if you want.
workspace;  % Make sure the workspace panel is showing.
format long g;
format compact;
fontSize = 15;
% Specify the folder where the files live.
inputImagesFolder = 'D:\whatever'; % Change it to whatever you need, if you want/need to.
% Check to make sure that folder actually exists.  Warn user if it doesn't.
if ~isfolder(inputImagesFolder)
	errorMessage = sprintf('Error: The following folder does not exist:\n\n%s\n\nPlease specify to an existing folder.', inputImagesFolder);
	uiwait(warndlg(errorMessage));
	% Try to find the highest level folder in that path that DOES exist.
	while ~isfolder(inputImagesFolder) && length(inputImagesFolder) > 3
		[inputImagesFolder, ~, ~] = fileparts(inputImagesFolder);
	end
	% Should have a good starting folder now.
	inputImagesFolder = uigetdir(inputImagesFolder);
	if inputImagesFolder == 0
		return;
	end
end
% Create output folder
outputImagesFolder = fullfile(inputImagesFolder, '/Resized to 224x224');
if ~isfolder(outputImagesFolder)
	mkdir(outputImagesFolder);
end
% Get a list of all files in the folder with the desired file name pattern.
filePattern = fullfile(inputImagesFolder, '*.jpg'); % Change to whatever pattern you need.
theFiles = dir(filePattern)
numFiles = length(theFiles);
hFig = figure;
hFig.WindowState = 'maximized';
for k = 1 : numFiles
	% Get the input file name.
	baseFileName = theFiles(k).name;
	fullFileName = fullfile(inputImagesFolder, baseFileName);
	% Get the output file name.
	outputFullFileName = fullfile(outputImagesFolder, baseFileName);
	fprintf(1, 'Now reading %d of %d "%s"\n', k, numFiles, fullFileName);
	
	% Read in input image with imread().
	inputImage = imread(fullFileName);
	% Resize it.
	outputImage = imresize(inputImage, [224, 224]);
	
	% Display input and output images.
	subplot(1, 2, 1);
	imshow(inputImage);  % Display image.
	caption = sprintf('Input image (%d of %d):\n"%s", %d by %d', k, numFiles, baseFileName, size(inputImage, 1), size(inputImage, 2));
	title(caption, 'FontSize', fontSize, 'Interpreter', 'none');
	subplot(1, 2, 2);
	imshow(outputImage);  % Display image.
	caption = sprintf('Output image: "%s", %d by %d', baseFileName, size(outputImage, 1), size(outputImage, 2));
	title(caption, 'FontSize', fontSize, 'Interpreter', 'none');
	drawnow; % Force display to update immediately.
	
	% Write out the resized output file to the output folder.
	imwrite(outputImage, outputFullFileName);
end
fprintf('Done running %s.m.\n', mfilename);
% Open the output folder in File Explorer.
winopen(outputImagesFolder);

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

The training images are of size 2062x3545x3 but the input layer expects images of size 32x32x3.

3 个评论
显示 1更早的评论隐藏 1更早的评论

回答（1 个）

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

另请参阅

类别

标签

Community Treasure Hunt

The training images are of size 2062x3545x3 but the input layer expects images of size 32x32x3.

3 个评论 显示 1更早的评论隐藏 1更早的评论

回答（1 个）

0 个评论 显示 -2更早的评论隐藏 -2更早的评论

另请参阅

类别

标签

Community Treasure Hunt

3 个评论
显示 1更早的评论隐藏 1更早的评论

0 个评论
显示 -2更早的评论隐藏 -2更早的评论