My Own Cascade Classifier: Can not get new positive sample. The most possible reason is insufficient count of samples in given vec-file.

asked 2017-10-22 04:48:05 -0500

wpadka gravatar image

updated 2017-10-22 05:26:08 -0500

berak gravatar image

Hi! I want to create my own cascade classifier to detect windows.Im using Cascade Trainger Gui. I have 2805 positive images and 2718 negatives, but when i try to build my classifier then ive got this error:

Can not get new positive sample. The most possible reason is insufficient count of samples in given vec-file.

I decreased my positive Image usage to 10 and this isnt still working and min hit rate to 0,5.

edit retag flag offensive close merge delete

Comments

what are you trying to train on (your object) ?

can you show us the cmdline params / logs ?

how did you get the data (annotated info file) ?

berak gravatar imageberak ( 2017-10-22 06:29:29 -0500 )edit

I want to detect windows in buildings. Here is link to my Cascade folder with all logs and pics: link text

wpadka gravatar imagewpadka ( 2017-10-22 10:17:26 -0500 )edit

Someone know how to do this?

wpadka gravatar imagewpadka ( 2017-10-25 08:32:24 -0500 )edit