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

Fig. 7.

Fig. 7

Fitting performance of SIFT-based AAMs and GN-DPMs on LFPW, Helen and AFW: Average pt-pt Euclidean error (normalized by the face size) Vs fraction of images. 68 points were used