Matlab AlexNet 识别花

1. 首先，你要又并行计算的工具箱，在插件选项里面找到，安装即可

2. 下载训练的数据集，采用matlab演示的材料即可

https://matlabacademy-content.mathworks.com/3.3/R2017b/content/deeplearning_course_files.zip

3. 运行训练脚本：

The code below implements transfer learning for the flower species example in this chapter. It is available as the script trainflowers.mlx in the course example files. You can download the course example files from the help menu in the top-right corner. Note that this example can take some time to run if you run it on a computer that does not have a GPU.

Get training images

flower_ds = imageDatastore('Flowers','IncludeSubfolders',true,'LabelSource','foldernames');
[trainImgs,testImgs] = splitEachLabel(flower_ds,0.6);
numClasses = numel(categories(flower_ds.Labels));

 

Create a network by modifying AlexNet

net = alexnet;
layers = net.Layers;
layers(end-2) = fullyConnectedLayer(numClasses);
layers(end) = classificationLayer;

 

Set training algorithm options

options = trainingOptions('sgdm','InitialLearnRate', 0.001);

 

Perform training

[flowernet,info] = trainNetwork(trainImgs, layers, options);

 

Use trained network to classify test images

testpreds = classify(flowernet,testImgs);

4. 运行报错，GPU内存不够

设置小一点：options = trainingOptions('sgdm','InitialLearnRate', 0.001,'MiniBatchSize', 64);

options = 


  TrainingOptionsSGDM - 属性:


                     Momentum: 0.9000
             InitialLearnRate: 1.0000e-03
    LearnRateScheduleSettings: [1×1 struct]
             L2Regularization: 1.0000e-04
      GradientThresholdMethod: 'l2norm'
            GradientThreshold: Inf
                    MaxEpochs: 30
                MiniBatchSize: 128
                      Verbose: 1
             VerboseFrequency: 50
               ValidationData: []
          ValidationFrequency: 50
           ValidationPatience: 5
                      Shuffle: 'once'
               CheckpointPath: ''
         ExecutionEnvironment: 'auto'
                   WorkerLoad: []
                    OutputFcn: []
                        Plots: 'none'
               SequenceLength: 'longest'

         SequencePaddingValue: 0

5. 结果