%定義efficientnet的結構 layers = [imageInputLayer([128 128 3]);%注意，128,128，3是訓練樣本的大小，這個和參考文獻不一樣，要根據實際輸入設置%stage1convolution2dLayer(floor(depth)+1,floor(width)+1,'Padding','same');%layer 1batchNormalizationLayer;reluLayer;maxPooling2dLayer(floor(resl)+1,'Stride',2);%stage2convolution2dLayer(floor(depth)+1,floor(width)+1,'Padding','same');%layer 1batchNormalizationLayer;reluLayer;%stage3convolution2dLayer(floor(depth)+1,floor(width)+1,'Padding','same');%layer 2batchNormalizationLayer;convolution2dLayer(floor(depth)+1,floor(width)+1,'Padding','same');%layer 2batchNormalizationLayer;reluLayer; %stage4convolution2dLayer(floor(depth)+1,floor(width)+1,'Padding','same');%layer 2batchNormalizationLayer;convolution2dLayer(floor(depth)+1,floor(width)+1,'Padding','same');%layer 2batchNormalizationLayer;reluLayer;maxPooling2dLayer(floor(resl)+1,'Stride',2);%stage5convolution2dLayer(floor(depth)+1,floor(width)+1,'Padding','same');%layer 3batchNormalizationLayer;convolution2dLayer(floor(depth)+1,floor(width)+1,'Padding','same');%layer 3batchNormalizationLayer;convolution2dLayer(floor(depth)+1,floor(width)+1,'Padding','same');%layer 3batchNormalizationLayer;reluLayer;maxPooling2dLayer(floor(resl)+1,'Stride',2);%stage6convolution2dLayer(floor(depth)+1,floor(width)+1,'Padding','same');%layer 3batchNormalizationLayer;convolution2dLayer(floor(depth)+1,floor(width)+1,'Padding','same');%layer 3batchNormalizationLayer;convolution2dLayer(floor(depth)+1,floor(width)+1,'Padding','same');%layer 3batchNormalizationLayer;reluLayer;%stage7convolution2dLayer(floor(depth)+1,floor(width)+1,'Padding','same');%layer 4batchNormalizationLayer;convolution2dLayer(floor(depth)+1,floor(width)+1,'Padding','same');%layer 4batchNormalizationLayer;convolution2dLayer(floor(depth)+1,floor(width)+1,'Padding','same');%layer 4batchNormalizationLayer;convolution2dLayer(floor(depth)+1,floor(width)+1,'Padding','same');%layer 4batchNormalizationLayer;reluLayer;maxPooling2dLayer(floor(resl)+1,'Stride',2);%stage8convolution2dLayer(floor(depth)+1,floor(width)+1,'Padding','same');%layer 1batchNormalizationLayer;reluLayer;maxPooling2dLayer(floor(resl)+1,'Stride',2);%stage9convolution2dLayer(floor(depth)+1,floor(width)+1,'Padding','same');%layer 1batchNormalizationLayer;reluLayer;%FCfullyConnectedLayer(CLASSNUM);%softmaxsoftmaxLayer;%輸出分類結果classificationLayer;];options = trainingOptions('sgdm', ...'InitialLearnRate', 0.01, ...'MaxEpochs', 200, ...'Shuffle', 'every-epoch', ...'ValidationData', imdsValidation, ...'ValidationFrequency', 5, ...'Verbose', false, ...'Plots', 'training-progress'); rng(1); %使用訓練集訓練網絡 net = trainNetwork(imdsTrain, layers, options);

訓練過程如下：

訓練精度為94.17%。

平均損失過程如下：

不同訓練樣本數量對應的訓練性能（注意，每次訓練會有一定的波動和偏差）

訓練樣本比例	改進前的訓練性能	改進后的訓練性能
5%	85.46%	92.23%
10%	89.20%	90.08%
20%	94.65%	92.94%
40%	93.53%	94.82%
60%	94.66%	98.06%
80%	94.67%	98.08%
90%	98.08%	100%

?A05-79

總結

以上是生活随笔為你收集整理的【Efficient-Net】基于Efficient-Net效滤网的目标识别算法的MATLAB仿真的全部內容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網站內容還不錯，歡迎將生活随笔推薦給好友。