============ SUPPLEMENTARY MATERIAL ==================
This supplementary material accompanies our submission "Pose Independent
Face Recognition by Localizing Local Binary Patterns via Deformation
Components". In particular we provide a link to a video hosted on Vimeo
that summarizes some computer animations that we could not add to the paper.
More about the video link at the end of this document.
The video shows the main building blocks of the proposed approach, that
are the following:
(A) 3D Pose Estimation
(B) Deformation Components via 3DMM
(C) Comparison: fitting a 3DMM without and with Regularization.
Please read the following before watching the video.
===== (A) 3D Pose Estimation ========
We show some qualitative examples about our 3D Pose Estimation
(Section IV-A of the paper) applied to another dataset. In particular
the video shows the estimate of pose applied to the 2D imagery of
Florence Faces dataset .
===== (B) Deformation Components via 3DMM ========
We show two animations about the Deformation Components learnt
via 3DMM, showing something similar to Fig. 1 of the paper but
animated. Firstly we show 19 components and then the 4th component
observed simultaneously from a frontal and lateral view. This component
affects nose deformation.
===== (C) Fitting a Regularized 3DMM ========
Finally we qualitatively compare the fitting of 3DMM without and
with the proposed regularization as Ridge regression solution (similar
to Fig. 2 of the paper). This part shows on the left the reprojection
error of detected landmarks (red) with the projected model landmarks
(magenta); on the right the 3D recovered shape.
===== VIDEO LINK ========
The video is publicly available at the following URL:
===== References ========