基于卷积神经网络CNN的图像识别，基于卷积神经网络的图像评价

clear clc close all rng(1) %% 1. 准备 imds = imageDatastore('CNN/', ... 'IncludeSubfolders',true, ... 'LabelSource','foldernames'); % 图像数据集 targets = xlsread("整合最终.xls") ; targets(121:122,:)=[]; [out,ps] = mapminmax(targets(:,end)'); targets = out; XTrain = zeros(74,144,3,100); XValidation = zeros(74,144,3,20); YTrain = zeros(100,1); YValidation = zeros(20,1); idx = randperm(120); t = 1; for i = idx img = im2double(imread(imds.Files{i})); % img = (img - min(img(:))) / (max(img(:)) - min(img(:))); if t <=100 % 训练集 XTrain(:,:,:,t) = img; YTrain(t) = targets(i); XValidation(:,:,:,t) = img; YValidation(t) = targets(i); else XValidation(:,:,:,t) = img; YValidation(t) = targets(i); end t = t+1; end %% 2.定义网络架构 layers = [ imageInputLayer([74 144 3],"Name","imageinput") convolution2dLayer([3 3],32,"Name","Conv2d_1","Padding","same") batchNormalizationLayer reluLayer maxPooling2dLayer(2,'Stride',2); % 37 72 convolution2dLayer([3 3],64,"Name","Conv2d_2","Padding","same") batchNormalizationLayer reluLayer maxPooling2dLayer(2,'Stride',2); % 18 36 convolution2dLayer([3 3],32,"Name","Conv2d_3","Padding","same") batchNormalizationLayer reluLayer maxPooling2dLayer(2,'Stride',2); % 9 18 fullyConnectedLayer(512,"Name","fc1") reluLayer dropoutLayer(0.09) fullyConnectedLayer(256,"Name","fc2") reluLayer fullyConnectedLayer(1) regressionLayer]; analyzeNetwork(layers) %% 3.训练网络 % 指定训练选项并训练网络。 options = trainingOptions('adam', ... 'MiniBatchSize', 2, ... 'MaxEpochs', 20, ... 'InitialLearnRate', 1e-5, ... 'LearnRateSchedule', 'piecewise', ... 'LearnRateDropFactor', 0.1, ... 'LearnRateDropPeriod', 20, ... 'Shuffle', 'every-epoch', ... 'Plots', 'training-progress', ... 'Verbose', true); net = trainNetwork(XTrain,YTrain, layers,options); %% 4.测试网络 YPredict = predict(net,XValidation); YPredict = mapminmax('reverse',YPredict',ps); YValidation = mapminmax('reverse',YValidation',ps); figure; plot(YPredict,'-o','LineWidth',1.5);hold on plot(YValidation,'--*','LineWidth',1.5); legend('预测','事实') [MSE, RMSE, MBE, MAE ] =MSE_RMSE_MBE_MAE(YValidation,YPredict) R2 = R_2(YValidation,YPredict) R2 = R_2(YValidation(101:end),YPredict(101:end))