How does facial landmarking work ?

asked 2015-03-23 08:50:20 -0500

tomnjerry gravatar image

updated 2015-04-20 08:44:15 -0500

While working on emotion detection, I came across a number of techniques such as using ASM, SIFT etc. All the techniques have assumed that I have all the landmark points (points on my eyes, nose, mouth etc) already detected. I am unable to figure out, how do I locate these points on image of face.

I can locate face and mark features such as eyes, mouth, nose using cascades but point landmarking (having points represent your eyes and so on) is something I am unable to understand. Any document to help me out with this!

edit retag flag offensive close merge delete

Comments

1

there are a few libraries for that, flandmark, dlib, also https://github.com/MasteringOpenCV/co...

berak gravatar imageberak ( 2015-03-23 08:59:09 -0500 )edit

Do this libraries directly help me mark the landmark points? I want to understand how does landmarking work basically and implement it myself if thats possible.

tomnjerry gravatar imagetomnjerry ( 2015-03-23 09:01:20 -0500 )edit
2

For example, check the thesis for flandmarks: ftp://cmp.felk.cvut.cz/pub/cmp/articles/uricar/Uricar-TR-2011-05.pdf Very well explained

albertofernandez gravatar imagealbertofernandez ( 2015-03-23 10:21:14 -0500 )edit