Conclusions
nThe
paper discussed the following three issues:
nLearning
class-specific information about human faces from a dataset of
examples.
nEstimating 3D shape and texture along with all relevant
3D scene parameters.
nRepresenting
and comparing faces for recognition tasks.