Ask Your Question
0

alignment of face data needed for both train and probe?

asked 2016-09-20 16:43:39 -0500

atv gravatar image

If i align my dataset for training..do i need to align the faces i get from my webcam as well ? I don't think philip's example shows that although i might be wrong.

I mean if we preprocess training data we need to do the same for the images we get from our webcam, so does this apply to alignment as well?

Thanks,

edit retag flag offensive close merge delete

Comments

It's not really clear from the question which OpenCV function(s) you are talking about. You might want to edit the question with a link to the function documentation/example you're talking about.

AJW gravatar imageAJW ( 2016-09-21 03:46:16 -0500 )edit

http://bytefish.de/blog/aligning_face...

http://www.openu.ac.il/home/hassner/p...

Not so much a function as more of an approach to align facial data using the eyes as an offset.

atv gravatar imageatv ( 2016-09-21 03:54:47 -0500 )edit

1 answer

Sort by ยป oldest newest most voted
0

answered 2016-09-22 06:32:43 -0500

YES, you need to apply the same operations on your test data as those you applied on your training data to get the full potential of your classifier. If not it will probably work but just less accurate.

edit flag offensive delete link more

Comments

Thanks Steven, that is very helpful

atv gravatar imageatv ( 2016-09-22 09:51:42 -0500 )edit

Question Tools

1 follower

Stats

Asked: 2016-09-20 16:43:39 -0500

Seen: 68 times

Last updated: Sep 20 '16