Skip to main content
. 2016 Sep 24;122(1):17–33. doi: 10.1007/s11263-016-0950-1

Fig. 1.

Fig. 1

First row: face images taken from the test set of LFPW (Belhumeur et al. 2011). The images were not seen during training. Second row: reconstruction of the images from the appearance subspace. The appearance subspace is powerful because the AAM was built in the wild