Data auguamtion in Matlab

To enlarge the dataset , I test the function of DA method with Xflection and Yflection. The first quesiton how to apply DA method.

Let's see the code.

randomAffine2d ：When XReflection is true (1), the transformation tform reflects images horizontally with 50% probability. By default, the transformation does not reflect images in the horizontal direction.

But I want 100% probability, how to realize this goal?

I have Edit the original file, and you can download it from my resouce, it will has 100% xReflection or YRelfection.

download  function

1 Read the image and label 

%1  read file  and label
I=imread(trainingDataTbl.imageFilename{1}); %% GET IMAGE
%B = cell(size(A));
BoxA=trainingDataTbl.vehicle{1} % GET LABEL
%I = A{1};
sz = size(I)
% if numel(sz)==3 && sz(3) == 3
%     I = jitterColorHSV(I,...
%         'Contrast',0.2,...
%         'Hue',0,...
%         'Saturation',0.1,...
%         'Brightness',0.2);
% end

    annotatedI = insertShape(I,'Rectangle',BoxA);
figure, imshow(annotatedI)
title('Original data')

2 Flect  the image and its coresponding label

% 2  Randomly flip and scale image.
tform = randomAffine2d('XReflection',true,'Scale',[1 1]); %X flection
rout = affineOutputView(sz,tform,'BoundsStyle','CenterOutput');
C = imwarp(I,tform,'OutputView',rout);
figure,imshow(C)
imwrite(C,'xflextion.jpg')
title('Flip data')

% 3 flip label
[D,indices] = bboxwarp(BoxA,tform,rout,'OverlapThreshold',0.25);
    annotatedJ = insertShape(C,'Rectangle',D);
figure, imshow(annotatedJ)
title('Flip data')
% 4test for write image and read again
E=imread('xflextion.jpg');
 size(E)

  annotatedJ = insertShape(E,'Rectangle',D);
figure, imshow(annotatedJ)
title('Flip data from read data')

3 write the flected image and read again

% 4test for write image and read again
E=imread('xflextion.jpg');
 size(E)

  annotatedJ = insertShape(E,'Rectangle',D);
figure, imshow(annotatedJ)
title('Flip data from read data')

 XReflection or Y Refection  with 100% prabobility

download  function

call the function with DA method

%test yReflection
%trainingDataTbl=gTruth(1:10,:);
%YReflection
Action='YReflection'
strpath='G:\DA2021\YFlection'
YFlipGrth = DAFlipAll(gTruth,Action,strpath);

%XReflection
Action='XReflection'
strpath='G:\DA2021\XFlection'
XFlipGrth = DAFlipAll(gTruth,Action,strpath);

enlarge the dataset

%% concat with DA dataset
            gTruthDA=vertcat(gTruth,YFlipGrth,XFlipGrth);
            size(gTruthDA)

%% 数据集DA
rng(0);

shuffledIndices = randperm(height(gTruthDA));
idx = floor(0.6 * length(shuffledIndices) );

trainingIdx = 1:idx;
trainingDataTbl = gTruthDA(shuffledIndices(trainingIdx),:);

validationIdx = idx+1 : idx + 1 + floor(0.1 * length(shuffledIndices) );
validationDataTbl = gTruthDA(shuffledIndices(validationIdx),:);

testIdx = validationIdx(end)+1 : length(shuffledIndices);
testDataTbl = gTruthDA(shuffledIndices(testIdx),:);