How to calculate true positive , true negative, false positive and false negative as we have segmented and ground truth

calculate true positive , true negative, false positive and false negative as we have segmented and ground truth is that code is correct idx = (expected()==1)

p = length( expected(idx)) n = length( expected(~idx)) N = p+n tp = sum( expected(idx)== predicted(idx)) tn = sum( expected(~idx)== predicted(~idx)) fp = n-tn fn = p-tp

 accuracy=(tp+tn)/(tp+tn+fp+fn)

Answer 1

[KALYAN ACHARJYA]

%Last year I answered this way, you can avoid the loop (Recommended)

   TP=0;FP=0;TN=0;FN=0;
      for i=1:400;
          for j=1:400;
              if(gold_data(i,j)==1 & test_data(i,j)==1);
                  TP=TP+1;
              elseif(gold_data(i,j)==0 & test_data(i,j)==1);
                  FP=FP+1;
              elseif(gold_data(i,j)==0 & test_data(i,j)==0);
                  TN=TN+1;
              else
                  FN=FN+1;
              end
          end
      end

[marwa Saad]

thanks i will check the code

[Cretu Ioana]

Coul you tell me who's the gold data and test data? what it means, bacause I also want to calculete FN and TN and I have different notation? thanks

[Arslan Ahmed Awan]

gold_data is your original ground-truth image & test_data will be your segmented-image using your proposed method.

[Sameema Tariq]

Using these tpr and fpr, how will you calculate Reciever Operating Characteristics (ROC)? As ROC is a graph between tpr and fpr on different threshold values, and how will you use it?

[Fadi Alsuhimat]

Sameema Tariq

I think you want ROC to find EER??

if that true you have to graph ROC between FAR & FRR, as

[Asha D L]

400 means total number of samples?? gold_data (is it dataset?)

[Amel Kaouane]

thank you so much