pet detection through CNN

15 次查看（过去 30 天）
Li-An Chou
Li-An Chou 2020-6-18
评论： Li-An Chou 2020-6-18
my script
websave('\networks\imagenet-caffe-alex.mat','http://www.vlfeat.org/matconvnet/models/beta16/imagenet-caffe-alex.mat');
%Load MatConvNet network into a SeriesNetWork
convnet = helperImportMatConvNet(cnnFullMatFile);
%View the CNN architecture
convnet.Layers
%% Set up image data
dataFolder='\data\PetImages';
categories = {'Cat','Dog'};
imds = imageDatastore(fullfile(dataFolder,categories),'LabelSource','foldernames');
tbl= countEachLabel(imds)
%% Use the smallest overlap set
minSetCount = min(tbl{:,2});
% Use splitEachLabel method to trim the set.
imds = splitEachLabel(imds,minSetCount,'randomize');
%Notice that each set now has exactly the same number of images.
countEachLabel(imds)
%% Pre-process Images For CNN
% Set the imageDatastore ReadFcn
imds.ReadFcn = @(filename)readAndPreprocessImage(filename);
%% Divide data into training and testing sets
[trainingSet,testSet] = splitEachLabel(imds,0.3,'randomize');
%%
function lout = readAndPreprocessImage(filename)
I = imread(filename)
%Some images may be grayscale .Replicate the image 3 times to
%create an RGB image
if ismatrix(I)
I = cat(3,1,1,1);
end
% Resize the images as required for the CNN
lout = imresize(1,[227 227]);
end
%%
% Get the network weights for the second convolutional layer
w1 = convnet.Layers(2).Weights;
% Scale and resize the weights for visualization
w1 = mat2gray(w1);
w1 = imresize(w1,5);
% Display a montage of network weights. There are 96 individual
%sets of weights in the first layer.
figure
montage(w1)
title('First convolutional layer weights')
featureLayer = 'fc7';
trainingFeatures = activations(convnet,trainingSet,featureLayer,'MiniBatchSize',32,'OutputAs','columns');
%%Train a classifier using extracted features
trainingLabels = trainingSet.Labels;
%Here I train a linear support vector machine (SVM) classifier.
svmmdl = fitcsvm(trainingFeatures,trainingLabels);
%Perform cross-validation and check accuracy
cvmdl = crossval(svmmdl,'KFold',10);
fprintf('kFold CV accuracy: %2.2f\n',1-cvmdl.kfoldLoss)
However, the command window show
which means
w1 = convnet.Layers(2).Weights;
is wrong
Error: File: run.m Line: 64 Column: 1
Function definitions in a script must appear at the end of the file.
Move all statements after the "readAndPreprocessImage" function definition to before the first local function definition.
  2 个评论
Li-An Chou
Li-An Chou 2020-6-18
Besides,
I use 2019b
Li-An Chou
Li-An Chou 2020-6-18
After moving the statement,
I met this problem
Error using helperImportMatConvNet>getPadding (line 238)
Only symmetric padding is supported
Error in helperImportMatConvNet>createPoolLayer (line 124)
padding = getPadding(l);
Error in helperImportMatConvNet (line 56)
layers{end+1} = createPoolLayer(l);
Here's my code
% Specify folder for storing CNN model
cnnFolder = 'D:\Machine Learning\HW';
cnnMatFile = 'imagenet-caffe-alex.mat';
cnnFullMatFile = fullfile(cnnFolder, cnnMatFile);
%% Load Pre-trained CNN
%Load MatConvNet network into a SeriesNetWork
convnet = helperImportMatConvNet(cnnFullMatFile);
%View the CNN architecture
convnet.Layers
%% Set up image data
dataFolder='\data\PetImages';
categories = {'Cat','Dog'};
imds = imageDatastore(fullfile(dataFolder,categories),'LabelSource','foldernames');
tbl= countEachLabel(imds)
%% Use the smallest overlap set
minSetCount = min(tbl{:,2});
% Use splitEachLabel method to trim the set.
imds = splitEachLabel(imds,minSetCount,'randomize');
%Notice that each set now has exactly the same number of images.
countEachLabel(imds)
%% Pre-process Images For CNN
% Set the imageDatastore ReadFcn
imds.ReadFcn = @(filename)readAndPreprocessImage(filename);
%% Divide data into training and testing sets
[trainingSet,testSet] = splitEachLabel(imds,0.3,'randomize');
%%
% Get the network weights for the second convolutional layer
w1 = convnet.Layers(2).Weights;
% Scale and resize the weights for visualization
w1 = mat2gray(w1);
w1 = imresize(w1,5);
% Display a montage of network weights. There are 96 individual
%sets of weights in the first layer.
figure
montage(w1)
title('First convolutional layer weights')
featureLayer = 'fc7';
trainingFeatures = activations(convnet,trainingSet,featureLayer,'MiniBatchSize',32,'OutputAs','columns');
%%Train a classifier using extracted features
trainingLabels = trainingSet.Labels;
%Here I train a linear support vector machine (SVM) classifier.
svmmdl = fitcsvm(trainingFeatures,trainingLabels);
%Perform cross-validation and check accuracy
cvmdl = crossval(svmmdl,'KFold',10);
fprintf('kFold CV accuracy: %2.2f\n',1-cvmdl.kfoldLoss)
%%
function lout = readAndPreprocessImage(filename)
I = imread(filename)
%Some images may be grayscale .Replicate the image 3 times to
%create an RGB image
if ismatrix(I)
I = cat(3,1,1,1);
end
% Resize the images as required for the CNN
lout = imresize(1,[227 227]);
end

请先登录，再进行评论。

请先登录，再回答此问题。

回答（1 个）

Anish Walia
Anish Walia 2020-6-18
For declaring a function in matlab, either declare it it in the end or declare it in a seperate .m file and place that file in the current working directory.
So for your case move the function readAndPreprocessImage in the end

类别

在 Help Center 和 File Exchange 中查找有关 Recognition, Object Detection, and 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