Plotting results of SVM on the trained images in Matlab

I am trying to find horizon line with hog feature extraction and svm training in sea images. The SVM training is done and it is working. I would like to show predicted labels on test images.

`% Bring predicted label to the horizon/background split

hPredict_label = Predict_label(1:length(hlabelTest));

bPredict_label = Predict_label(length(hlabelTest)+1:length(hlabelTest)+length(blabelTest));

% Vizualize predicted/classified data

for i=1:original_imagestest_db.Count

original_file = original_imagestest_db.ImageLocation{i};

figure(200)

x = imread(original_file);

x = imresize(x,[400,400]);

imshow(x);

hold on

num_hf = size(dataset.HorizonFeatsTest{i},1);

num_bf = size(dataset.BgFeatsTest{i},1);

for n=1:num_hf

plot(hxhy(:,1),'ro');

end

for n=1:num_bf

plot(bxby(:,1),'go');

end

end `

actually bxby and hxhy are the x and y coordinates of randomly chosen on the images but the problem is i should not plot them. because they are from training part. I wanna plot the labels and show correct or wrong. When I try to plot hPredict_label it gaves 1 1 1 1 coordinates because these are the 1 labels for horizon on the image.

Do you have any suggestions or any idea?

Thank you

Answer 1

[Image Analyst]

I can't reproduce your results or help you since I don't have your data or code up until then. But if you "wanna plot the labels and show correct or wrong" I suggest you use sprintf() to make up text strings for labels, and use text() to place those strings in the overlay above the image.

[Elf]

thank you for the answer. but how can i plot labels? for example i want to make a comparison. such as i need feature test location which is hfsTest in my code and the correct label is in my ground truth data which is hfeats, i wanna plot the location on the image (is it possible to do that from features?) then if it is correct i will plot in green if it's not a correct classification in red.