5、matlab实现
<pre><code>img_path = 'D:\work\jupyterNotebook\baidu_dianshi\data\class';
imds = imageDatastore(img_path, ...
'IncludeSubfolders',true,'LabelSource','foldernames');</code></pre>
<pre><code>figure;
perm = randperm(2000,20);
for i = 1:20
subplot(4,5,i);
imshow(imds.Files{perm(i)});
end
</code></pre>
<p><img src="https://www.showdoc.cc/server/api/common/visitfile/sign/4b852b95ee3fe56f3099b70975cf9004?showdoc=.jpg" alt="" /></p>
<pre><code>labelCount = countEachLabel(imds)</code></pre>
<pre><code>CITY 55
DESERT 636
FARMLAND 218
LAKE 171
MOUNTAIN 464
OCEAN 456 </code></pre>
<pre><code>img = readimage(imds,1);
size(img)</code></pre>
<pre><code>ans = 1×3
256 256 3</code></pre>
<pre><code>net = alexnet;
net.Layers</code></pre>
<pre><code>ans =
25x1 Layer array with layers:
1 'data' Image Input 227x227x3 images with 'zerocenter' normalization
2 'conv1' Convolution 96 11x11x3 convolutions with stride [4 4] and padding [0 0 0 0]
3 'relu1' ReLU ReLU
4 'norm1' Cross Channel Normalization cross channel normalization with 5 channels per element
5 'pool1' Max Pooling 3x3 max pooling with stride [2 2] and padding [0 0 0 0]
6 'conv2' Convolution 256 5x5x48 convolutions with stride [1 1] and padding [2 2 2 2]
7 'relu2' ReLU ReLU
8 'norm2' Cross Channel Normalization cross channel normalization with 5 channels per element
9 'pool2' Max Pooling 3x3 max pooling with stride [2 2] and padding [0 0 0 0]
10 'conv3' Convolution 384 3x3x256 convolutions with stride [1 1] and padding [1 1 1 1]
11 'relu3' ReLU ReLU
12 'conv4' Convolution 384 3x3x192 convolutions with stride [1 1] and padding [1 1 1 1]
13 'relu4' ReLU ReLU
14 'conv5' Convolution 256 3x3x192 convolutions with stride [1 1] and padding [1 1 1 1]
15 'relu5' ReLU ReLU
16 'pool5' Max Pooling 3x3 max pooling with stride [2 2] and padding [0 0 0 0]
17 'fc6' Fully Connected 4096 fully connected layer
18 'relu6' ReLU ReLU
19 'drop6' Dropout 50% dropout
20 'fc7' Fully Connected 4096 fully connected layer
21 'relu7' ReLU ReLU
22 'drop7' Dropout 50% dropout
23 'fc8' Fully Connected 1000 fully connected layer
24 'prob' Softmax softmax
25 'output' Classification Output crossentropyex with 'tench' and 999 other classes</code></pre>
<pre><code>layersTransfer = net.Layers(2:end-3);</code></pre>
<pre><code>layersTransfer =
21x1 Layer array with layers:
1 'conv1' Convolution 96 11x11x3 convolutions with stride [4 4] and padding [0 0 0 0]
2 'relu1' ReLU ReLU
3 'norm1' Cross Channel Normalization cross channel normalization with 5 channels per element
4 'pool1' Max Pooling 3x3 max pooling with stride [2 2] and padding [0 0 0 0]
5 'conv2' Convolution 256 5x5x48 convolutions with stride [1 1] and padding [2 2 2 2]
6 'relu2' ReLU ReLU
7 'norm2' Cross Channel Normalization cross channel normalization with 5 channels per element
8 'pool2' Max Pooling 3x3 max pooling with stride [2 2] and padding [0 0 0 0]
9 'conv3' Convolution 384 3x3x256 convolutions with stride [1 1] and padding [1 1 1 1]
10 'relu3' ReLU ReLU
11 'conv4' Convolution 384 3x3x192 convolutions with stride [1 1] and padding [1 1 1 1]
12 'relu4' ReLU ReLU
13 'conv5' Convolution 256 3x3x192 convolutions with stride [1 1] and padding [1 1 1 1]
14 'relu5' ReLU ReLU
15 'pool5' Max Pooling 3x3 max pooling with stride [2 2] and padding [0 0 0 0]
16 'fc6' Fully Connected 4096 fully connected layer
17 'relu6' ReLU ReLU
18 'drop6' Dropout 50% dropout
19 'fc7' Fully Connected 4096 fully connected layer
20 'relu7' ReLU ReLU
21 'drop7' Dropout 50% dropout</code></pre>
<pre><code>layers = [
imageInputLayer([256 256 3])
layersTransfer
fullyConnectedLayer(6,'WeightLearnRateFactor',20,'BiasLearnRateFactor',20)
softmaxLayer
classificationLayer];
</code></pre>
<pre><code>options = trainingOptions('sgdm', ...
'MiniBatchSize',10, ...
'MaxEpochs',6, ...
'InitialLearnRate',1e-4, ...
'ValidationData',imdsValidation, ...
'ValidationFrequency',3, ...
'ValidationPatience',Inf, ...
'Verbose',false, ...
'Plots','training-progress');</code></pre>
<pre><code>netTransfer = trainNetwork(imdsTrain,layers,options);</code></pre>
<p><img src="https://www.showdoc.cc/server/api/common/visitfile/sign/4cb77ae83ef568837d456d9c72564bcb?showdoc=.jpg" alt="" /></p>
<pre><code>YPred = classify(netTransfer,imdsValidation);
YValidation = imdsValidation.Labels;
accuracy = sum(YPred == YValidation)/numel(YValidation)</code></pre>
<p>accuracy = 0.9400</p>
<h5>训练完可以把所有变量保存到mat文件</h5>
<hr />
<hr />
<h5>测试时加载mat文件，对未知数据测试</h5>
<pre><code>YPred = classify(netTransfer,imdsValidation);
YValidation = imdsValidation.Labels;
accuracy = sum(YPred == YValidation)/numel(YValidation)</code></pre>
<p>accuracy = 0.9400</p>
<pre><code>pre_img_path = 'D:\work\jupyterNotebook\baidu_dianshi\data\pre_data';
pre_imds = imageDatastore(pre_img_path, ...
'IncludeSubfolders',true,'LabelSource','foldernames');</code></pre>
<pre><code>pre_Pred =
DESERT
OCEAN
MOUNTAIN
MOUNTAIN
LAKE
DESERT
LAKE
LAKE
...</code></pre>
<hr />
<p>导出变量</p>
<pre><code>pfiles = pre_imds.Files
</code></pre>
<pre><code>pfiles = {'D:\work\jupyterNotebook\baidu_dianshi\data\pre_data\MWI_KD96UmGQ6KWVeohF.jpg'}
{'D:\work\jupyterNotebook\baidu_dianshi\data\pre_data\MWI_KDDTIKoSiQgwZiUQ.jpg'}
{'D:\work\jupyterNotebook\baidu_dianshi\data\pre_data\MWI_KERiJD55HvBKIhmL.jpg'}
{'D:\work\jupyterNotebook\baidu_dianshi\data\pre_data\MWI_KF9KpqQNNMVSsUKH.jpg'}
....</code></pre>
<pre><code>save pfile.mat pfiles</code></pre>
<pre><code>% cellstr 直接转为cell array类型 便于导入python
cell_pre = cellstr(pre_Pred);
save cell_str.mat cell_pre
</code></pre>