Ask Your Question

Revision history [back]

click to hide/show revision 1
initial version

Train Haar classifier for open/closed eyes


I am working in a Haar classifier for open/closed eyes. So far I have 700 unique positive images and 4000 negative images. From the 700 positive images I create 15000 changing rotation, brightness and contrast. I think this amount of images is enough for a good classifier, am I right?

The main problem is that I am not getting the expected results as I dont really understand how the parameters work. Which is the relation between the number of positive and negative images? How is that also related to the number of stages?

By the way, does anybody know about some good tutorial on this?? Really appreciated any help.