Can you trick a facial recognition Time Clock with a photo print?

AlphaBiz Solutions
17 July, 2017

We get a lot of questions from customers regarding how the FingerTec’s bio-metrics work and today I have picked out another specific question that I will address, and remember there is no wrong question, so keep them coming. If you would like to see our last question “Can the fingerprint data be used for other criminal purposes?” please click here.

Otherwise, here is today’s question “Can I print a picture of my face to trick the FaceID2?”

Like I said there is no wrong question, and so this seems pretty logical if a device recognizes a face why can’t I use a picture, even a high-quality photo print. Let’s see if we can break it down, first of all, what is the difference between a photograph and a real-life person, let me list a few differences here.

  • The picture has only one angle.
  • The picture is flat.
  • The picture is static.

As you can see they are obvious differences but only one has a real effect on a facial recognition device like the FingerTec FaceID2 and that is the flatness of the picture. If the FacID2 took only your picture then this method would work, but the FaceID2 doesn’t do that for authentication. The only picture is there for ascetics.

Let me explain the hardware behind the device and why it doesn’t work with a picture. The FaceID2 has two cameras next to each other, like our eyes. This way the device can sense ‘depth of field’, which is the distance between two points of reference. This means the camera can tell if the image in front of it is flat or three-dimensional and that is simply why it will not work with a printed image.

Let’s not stop there, however, lets answer, ‘how does the FaceID2 recognize your face?’ The device now sees your face in three-dimensions but it is not a sentient being and cannot tell that this face belongs to you, instead of in an instant, and I mean it’s really quick, it finds natural reference points on your face. Just like in the spy movies, each point represents a unique measurement of your face, such as cheekbones, eye level, basically your general face structure.

All this is converted to ones and zeros no actual image is used. Like in our previous blog it matches the raw data against what’s stored in its encrypted memory and if it matches the time and attendance is recorded.