tri_68.mat file. about clm-framework HOT 3 CLOSED

It was created using a Delaunay triangulation (Matlab has a function to do this) of landmark points of the mean shape from the automatically constructed PDM (using only the X and Y components). The result was later manually post-processed to cut out mouth and eye regions. Hope this answers your question. Let me know if you have any more questions, but possibly the best way to do this is by e-mailing me rather than raising an issue, as this is not really a problem or a bug with the code.

from clm-framework.

What is the rationale behind cutting out the mouth and eye regions?

from clm-framework.

The reason is to make avatar animation or expression transfer easier, for example for projects like this - http://www.cl.cam.ac.uk/~tb346/pub/papers/idgei2014.pdf, and is used more in the FAME branch of the code.

It doesn't have much bearing on landmark detection, but for simplicity the same triangulation is used in face validation stage.

from clm-framework.