How to solve this error cause our origin pic is 512512 but my GPU is not good enough so training is too slow and i wanna use 256256 to let it train faster!

Question

FangMing Tsao 2022-6-3

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1732990-how-to-solve-this-error-cause-our-origin-pic-is-512-512-but-my-gpu-is-not-good-enough-so-training-is

回答： Dinesh 2023-3-2

imds1 = imageDatastore('E:\Picture\PDRR');

imds2 = imageDatastore ('E:\Picture\DRR');

augmenter = imageDataAugmenter( ...

'RandRotation',@()randi([0,1],1)*90, ...

'RandXReflection',true);

miniBatchSize = 4;

patchSize = [256 256];

patchds = randomPatchExtractionDatastore(imds1,imds2,patchSize, ...

'PatchesPerImage',8, ...

'DataAugmentation',augmenter);

patchds.MiniBatchSize = miniBatchSize;

lgraph=layerGraph();

tempLayers=[

imageInputLayer([256 256 1],"Name","ImageInputLayer","Normalization","rescale-zero-one")

convolution2dLayer([3 3],32,"Name","Encoder-Stage-1-Conv-1","Padding","same","WeightsInitializer","he")

reluLayer("Name","Encoder-Stage-1-ReLU-1")

convolution2dLayer([3 3],32,"Name","Encoder-Stage-1-Conv-2","Padding","same","WeightsInitializer","he")

reluLayer("Name","Encoder-Stage-1-ReLU-2")];

lgraph=addLayers(lgraph,tempLayers);

tempLayers=[

maxPooling2dLayer([2 2],"Name","Encoder-Stage-1-MaxPool","Stride",2 )

convolution2dLayer([3 3],64,"Name","Encoder-Stage-2-Conv-1","Padding","same","WeightsInitializer","he")

reluLayer("Name","Encoder-Stage-2-ReLU-1")

convolution2dLayer([3 3],64,"Name","Encoder-Stage-2-Conv-2","Padding","same","WeightsInitializer","he")

reluLayer("Name","Encoder-Stage-2-ReLU-2")];

lgraph=addLayers(lgraph,tempLayers);

tempLayers=[

maxPooling2dLayer([2 2],"Name","Encoder-Stage-2-MaxPool","Stride",2)

convolution2dLayer([3 3],128,"Name","Encoder-Stage-3-Conv-1","Padding","same","WeightsInitializer","he")

reluLayer("Name","Encoder-Stage-3-ReLU-1")

convolution2dLayer([3 3],128,"Name","Encoder-Stage-3-Conv-2","Padding","same","WeightsInitializer","he")

reluLayer("Name","Encoder-Stage-3-ReLU-2")];

lgraph=addLayers(lgraph,tempLayers);

tempLayers=[

dropoutLayer(0.5,"Name","Encoder-Stage-3-DropOut")

maxPooling2dLayer([2 2],"Name","Encoder-Stage-3-MaxPool","Stride",2)

convolution2dLayer([3 3],256,"Name","Bridge-Conv-1","Padding","same","WeightsInitializer","he")

reluLayer("Name","Bridge-ReLU-1")

convolution2dLayer([3 3],256,"Name","Bridge-Conv-2","Padding","same","WeightsInitializer","he")

reluLayer("Name","Bridge-ReLU-2")

dropoutLayer(0.5,"Name","Bridge-DropOut")

transposedConv2dLayer([2 2],128,"Name","Decoder-Stage-1-UpConv","Stride",2)

reluLayer("Name","Decoder-Stage-1-UpReLU")];

lgraph=addLayers(lgraph,tempLayers);

tempLayers=[

depthConcatenationLayer(2,"Name","Decoder-Stage-1-DepthConcatenation")

convolution2dLayer([3 3],128,"Name","Decoder-Stage-1-Conv-1","Padding","same","WeightsInitializer","he")

reluLayer("Name","Decoder-Stage-1-ReLU-1")

convolution2dLayer([3 3],128,"Name","Decoder-Stage-1-Conv-2","Padding","same","WeightsInitializer","he")

reluLayer("Name","Decoder-Stage-1-ReLU-2")

transposedConv2dLayer([2 2],64 ,"Name","Decoder-Stage-2-UpConv","Stride",2)

reluLayer("Name","Decoder-Stage-2-UpReLU")];

lgraph=addLayers(lgraph,tempLayers);

tempLayers=[

depthConcatenationLayer(2,"Name","Decoder-Stage-2-DepthConcatenation")

convolution2dLayer([3 3],64,"Name","Decoder-Stage-2-Conv-1","Padding","same","WeightsInitializer","he")

reluLayer("Name","Decoder-Stage-2-ReLU-1")

convolution2dLayer([3 3],64,"Name","Decoder-Stage-2-Conv-2","Padding","same","WeightsInitializer","he")

reluLayer("Name","Decoder-Stage-2-ReLU-2")

transposedConv2dLayer([2 2],32,"Name","Decoder-Stage-3-UpConv","Stride",2)

reluLayer("Name","Decoder-Stage-3-UpReLU")];

lgraph=addLayers(lgraph,tempLayers);

tempLayers=[

depthConcatenationLayer(2,"Name","Decoder-Stage-3-DepthConcatenation")

convolution2dLayer([3 3],32,"Name","Decoder-Stage-3-Conv-1","Padding","same","WeightsInitializer","he")

reluLayer("Name","Decoder-Stage-3-ReLU-1")

convolution2dLayer([3 3],32,"Name","Decoder-Stage-3-Conv-2","Padding","same","WeightsInitializer","he")

reluLayer("Name","Decoder-Stage-3-ReLU-2")

convolution2dLayer([2 2],1,"Name","Final-ConvolutionLayer","Padding","same","WeightsInitializer","he")

regressionLayer('Name','output')

];

lgraph=addLayers(lgraph,tempLayers);

clear tempLayers;

lgraph=connectLayers(lgraph,"Encoder-Stage-1-ReLU-2","Encoder-Stage-1-MaxPool");

lgraph=connectLayers(lgraph,"Encoder-Stage-1-ReLU-2","Decoder-Stage-3-DepthConcatenation/in2");

lgraph=connectLayers(lgraph,"Encoder-Stage-2-ReLU-2","Encoder-Stage-2-MaxPool");

lgraph=connectLayers(lgraph,"Encoder-Stage-2-ReLU-2","Decoder-Stage-2-DepthConcatenation/in2");

lgraph=connectLayers(lgraph,"Encoder-Stage-3-ReLU-2","Encoder-Stage-3-DropOut");

lgraph=connectLayers(lgraph,"Encoder-Stage-3-ReLU-2","Decoder-Stage-1-DepthConcatenation/in2");

lgraph=connectLayers(lgraph,"Decoder-Stage-1-UpReLU","Decoder-Stage-1-DepthConcatenation/in1");

lgraph=connectLayers(lgraph,"Decoder-Stage-2-UpReLU","Decoder-Stage-2-DepthConcatenation/in1");

lgraph=connectLayers(lgraph,"Decoder-Stage-3-UpReLU","Decoder-Stage-3-DepthConcatenation/in1");

figure

plot(lgraph)

maxEpochs =1;

epochIntervals = 1;

initLearningRate = 0.01;

learningRateFactor = 0.0001;

l2reg = 0.0001;

options = trainingOptions('sgdm', ...

'Momentum',0.9, ...

'InitialLearnRate',initLearningRate, ...

'LearnRateSchedule','piecewise', ...

'LearnRateDropPeriod',10, ...

'LearnRateDropFactor',learningRateFactor, ...

'L2Regularization',l2reg, ...

'MaxEpochs',maxEpochs ,...

'MiniBatchSize',miniBatchSize, ...

'GradientThresholdMethod','l2norm', ...

'Plots','training-progress', ...

'GradientThreshold',0.01);

modelDateTime = datestr(now,'dd-mmm-yyyy-HH-MM-SS');

net = trainNetwork(patchds,lgraph,options);

save(['trainedNet-' modelDateTime '-Epoch-' num2str(maxEpochs*epochIntervals) ...

'ScaleFactors-' num2str(234) '.mat'],'net','options');

pic = imread('E:\Picture\PDRR\00001.jpg');

out2 = predict(net,pic);

subplot(1,2,1)

imshow(pic)

subplot(1,2,2)

imshow(out2,[])

2 个评论
显示无隐藏无

Seth Furman 2022-6-6

编辑：Seth Furman 2022-6-6

在 MATLAB Online 中打开

I should mention that datestr is discouraged. Prefer datetime where possible.

https://www.mathworks.com/help/matlab/ref/datetime.html

For example,

dt = datetime("now","Format","dd-MMM-yyyy-HH-mm-ss")
dt = datetime
   06-Jun-2022-16-07-40
string(dt)
ans = "06-Jun-2022-16-07-40"

Joss Knight 2022-6-11

Is randomPatchExtractionDatastore what you actually want, or is it just your attempt to reduce the input size?

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Dinesh 2023-3-2

Hi FangMing!

As per my understanding you want to resize the picture from 512*512 to 256 * 256 to increase the training speed.

You can use the function ‘imresize’ to resize the image, set the ‘scale’ parameter to 0.5.

You can refer the following MATLAB documentation for more details:

Hope this helps !

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

How to solve this error cause our origin pic is 512512 but my GPU is not good enough so training is too slow and i wanna use 256256 to let it train faster!

2 个评论
显示无隐藏无

回答（1 个）

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

另请参阅

类别

标签

产品

版本

Community Treasure Hunt

How to solve this error cause our origin pic is 512*512 but my GPU is not good enough so training is too slow and i wanna use 256*256 to let it train faster!

2 个评论 显示 无隐藏 无

回答（1 个）

0 个评论 显示 -2更早的评论隐藏 -2更早的评论

另请参阅

类别

标签

产品

版本

Community Treasure Hunt

WeChat

How to solve this error cause our origin pic is 512512 but my GPU is not good enough so training is too slow and i wanna use 256256 to let it train faster!

2 个评论
显示无隐藏无

0 个评论
显示 -2更早的评论隐藏 -2更早的评论