Ask Your Question

How facemarkLBF work?

asked 2018-12-06 07:45:08 -0500

Sartogersofortotropher gravatar image

How facemarkLBF work?

edit retag flag offensive close merge delete

2 answers

Sort by ยป oldest newest most voted

answered 2018-12-06 08:25:27 -0500

phillity gravatar image

Here is the 2014 CVPR paper it is based on "Face alignment at 3000 fps via regressing local binary features"

edit flag offensive delete link more


@phillity In facemarkLBF cascade classifier is used to detect face. When res10_300x300_ssd_iter_140000 is used to detect face Facemark are not at same location. I don't think that's a problem. Are you agree with me?

LBerger gravatar imageLBerger ( 2019-07-27 15:06:06 -0500 )edit

@LBerger Ya, unless the facial landmark points seem wrong or are really different, I would think it's okay

The facial bounding boxes returned by the cascade classifier and the ssd will likely be different, so I would guess the facial landmarks extracted from each would also be different. Hopefully they would be close though

phillity gravatar imagephillity ( 2019-07-28 14:42:21 -0500 )edit

in green cascade classifier and in yellow res10_300x300_ssd_iter_140000 facemark are differents but the logic is good :

image description

LBerger gravatar imageLBerger ( 2019-07-28 15:19:32 -0500 )edit

Ya I think the logic is correct. Since the facial bounding boxes are different, facial landmarks are found using different regions of the image

See line 451

phillity gravatar imagephillity ( 2019-07-28 15:42:59 -0500 )edit

Question Tools

1 follower


Asked: 2018-12-06 07:44:44 -0500

Seen: 255 times

Last updated: Dec 06 '18