|
I have a task that detecting facial expressions. What are considered the best current features? What are the best papers and conferences to read to find recent work? |
|
I think SIFT (and related) features might be a good start for that task, as SIFT is essentially a glorified corner detector. Maybe you should also consider the relative position of the sift features as well, as they might be good clues that the corner of the mouth is raised in relatio to the eyes, or something. |
|
I found this paper to be a reasonable starting point. It suggests that specific facial deformations from a neutral face emotion serve as reasonable well performing features. These include movements of the eyebrows, shift of the gaze, and corners of the lips. @zaxfax Please don't answer your own question. Edit your question to include this suggestion in it. Or add it as a comment to the question.
(Dec 09 '10 at 15:36)
Aman
|
|
There are some work in trying to recognize a set of facial expression units (Defined by Ekman), since by detecting those you are basically able to infer any possible expression and feeling in the face. Would you be able to include a link to the paper you are referencing?
(Dec 09 '10 at 13:31)
zaxtax ♦
Sure, I was in my mobile, so I had no quick access to it: Ekman's paper on facial units is this: http://www.paulekman.com/wp-content/uploads/2009/02/Facial-Sign-Of-Emotional-Experience.pdf And this one is a good survey on the topic http://portal.acm.org/citation.cfm?id=1454711.1454712
(Dec 09 '10 at 15:31)
Leon Palafox ♦
Thanks for the link!
(Dec 09 '10 at 16:32)
zaxtax ♦
|
|
There are some work in trying to recognize a set of facial expression units (Defined by Ekman), since by detecting those you are basically able to infer any possible expression and feeling in the face. |
