机器学习（基于MATLAB的Alexnet迁移学习进行猫狗分类）

敏英玮 2022-09-25 13:24:03

习题案例源自作者：https://blog.csdn.net/garker?type=blog

软件：matlab2022a

数据集：kaggle猫狗数据集

网络：Alexnet

数据集准备：【数据集度云链接】
链接：https://pan.baidu.com/s/17c4K04kDKDUsuXdLkPecKA
提取码：8rhn

（D：/machine learning/Kaggle/test1&train）

说明：获取数据集后对其进行分类（训练集、测试集），对图片大小调整后，将训练集按照7:3进行分割，而后读取网络alexnet并组建新网络，然后训练、测试alexnet网络

一、打开matlab点击新建，新建后打开文件夹D：/machine learning/Kaggle/test1&train，代码放入后点击运行，因为数据集较大，所以运行时间较长。

二、完整代码

imds = imageDatastore('D:\machine learning\Kaggle\train', ...
    'IncludeSubfolders',true, ...
    'LabelSource','foldernames');

numTrainImages = numel(imds.Labels);

for i = 1:numTrainImages
    s = string(imds.Files(i));
    I = imread(s);
    I = imresize(I,[227,227]);
    imwrite(I,s);
    s
end

[imdsTrain,imdsValidation] = splitEachLabel(imds,0.7,'randomized');

net = alexnet;

inputSize = net.Layers(1).InputSize

layersTransfer = net.Layers(1:end-3);

numClasses = numel(categories(imdsTrain.Labels));

layers = [
    layersTransfer
    fullyConnectedLayer(numClasses,'WeightLearnRateFactor',20,'BiasLearnRateFactor',20)
    softmaxLayer
    classificationLayer];

augimdsTrain = augmentedImageDatastore(inputSize(1:2),imdsValidation);
augimdsValidation = augmentedImageDatastore(inputSize(1:2),imdsValidation);
options = trainingOptions('sgdm', ...
    'MiniBatchSize',10, ...
    'MaxEpochs',10, ...
    'InitialLearnRate',1e-4, ...
    'Shuffle','every-epoch', ...
    'ValidationData',augimdsValidation, ...
    'ValidationFrequency',3, ...
    'Verbose',false, ...
    'Plots','training-progress');

netTransfer = trainNetwork(augimdsTrain,layers,options);

三、训练结果