Haar Detection and Performance

asked 2017-04-26 12:31:10 -0500

Aylee gravatar image

When i am testing my classifier the rate of object detection is low. Similarly, when i run performance, it has high missed object rate. While there is no false detection. Initially my classifier was not detecting at all. I changed half of my negative images (as previously i was using grey scale images, i replaced them with colorful ones), then it started detecting objects. Using grey scale images has something to do with it? (still half of my negative data is in grey scale) Thank you,

edit retag flag offensive close merge delete

Comments

all images will be converted internally to greyscale, so if your negatives are grey or not does never matter at all.

berak gravatar imageberak ( 2017-04-29 01:33:11 -0500 )edit