deepinsight
Inspireface liveness detection problem
I used inspireface python package to liveness detection.
Why rgb liveness confidence is incorrect? What is wrong?
Please help me
maybe because the liveness detection model crops and aligns the face first? if so then the only factors that can be used to determine if it's a flat picture is if the face/picture is not perfectly centered and not tilted on the Z axis because if you look down some facial features become occluded but if you tilt a flat picture it will look like the face is the same but there is a warp where closer things are bigger than further things the other factor is if there is a screen-door effect where pixel grids are obvious
so try showing the image on a smartphone screen or try tilting the top of the picture forwards or backwards and see if that works, if so then it's a limitation of the implementation because it's impossible to tell a fake printed picture from a tightly cropped image
