I have 1000 precision recall curves resulted from 1000 query images vs a database in Matlab. I wonder if anyone help to make 1000 curve interpolate and average to one precision recall curve. To plot each precision-recall curve I have 150 points for precision as a vector and the same number for recall.
Asked
Active
Viewed 74 times
0
-
What do your curves look like? How do you extract them from the images? You'll have to provide some code for people to work with... – lhcgeneva Aug 12 '15 at 13:47
-
Thank you for the link, the difference is that in that discussion the author address the problem with calculation of the precision and recall and plotting the curve for one query image. However, in my question we have 1000 query image so 1000 precision-recall curves (one for each query image) that need to be averaged to one curve inorder to show the overall performance of the system. – Hamid Aug 13 '15 at 14:19
-
my database has 2250 images from 15 classes, which mean each class of images has 150 images in the database. so based on this for each query image 150 top ranked images are selected and based on that precision and recall for each of query image is computed. so I pass 1000 query images that will produce 1000 precision-recall curves. it is needed to be interpolated and averaged to one curve. – Hamid Aug 13 '15 at 14:33