< Antispoofing > Does we should align face in Face Antispoofing

Align or not, Texture or shape

Relationship with Face Recognition(FR)

The aim of Face Recognition is to shorten the distance of one person from different scene, even from different medium. For example, FR should shorten the distance between the person in real life and the person’s picture in paper or tv or phone.

However, Face Antispoofing(FA) just did the opposite thing.

FA should shorten the person from the same medium. FA should shroten the distance between the two different real persons. And FA should widen the distance of the people from different medium, even the two persons are the same.
So as we can see, to differentiate medium is the key of FA.

What does the convlutional neural networks learn?

As we all know, FR should grab the features from two persons’ faces and differentiate them. The face texture(skin) a person can not be treated as a key feature because one person’s face texture may be different under different conditions. So FR should differentiate two persons’ face by their faces’ shape(My opinion).

Sadly, convlutional neural networks are good at grabbing the feature of image textrue, and do very poorly at grabbing the feature of shape.

So strictly aligment of the face is very import in FR. For FR, the aim of face alignment is to let the deep network to learn the difference of the face shape regardless of the face texture.

On the opposite, FA mainly learns the feature of texture of the medium regardless of the shape of ons’s face, which could take the advantage of learning capcity of the convlutional neural networks.

So alignment is not important in FA.