// ANTI–FACE

a wearable facial recognition destabilizer

We have put a demo of our ANTI–FACE technology in the signal chain, you must:

  1. allow the page to use your webcam
  2. make sure that your face is visible in the video, then click start
  3. keep your face still, and wait till the mask is applied
  4. try out different masks from the dropdown
This demo is a non-recording technology.

There was some problem trying to capture your webcamera, please check that your browser supports WebRTC. Using a fallback video instead. try it out:

  1. click start
  2. wait till model fits the face and a mask is applied
  3. try out different masks from the dropdown


Anti-Face explores the use of algorithms to defuse another algorithmic system. If a database of faces exists for comparison, how might we also take advantage of such a system to create recombinant shifting image identities?

We use an anti-face program to scramble or thwart the recognition algorithms at the input layer. This approach requires those who wish to use it to wear additional hardware but, finally, that hardware has become extremely lightweight.

The terrain of facial recognition is one where a politics of algorithms is most apparent—biometric algorithms embody normative conceptions of identity particularly as they are expressed in physical features.

A WEARABLE TECHNOLOGY

ANTI–FACE nano technology fits in any necklace or other near-face fashion or utilitarian items you may own. The optional subtle glow let's you know ANTI–FACE is on.

Similar to the Hövding invisible bicycle helmet, ANTI–FACE is always prepared to wrap you in its anti-face detection mesh.

OUR PROCESS

There are several distinct algorithmic approaches to facial recognition within images . One of the most well-known methods is to look for ‘landmarks’ on the face—defining features which can be compared and contrasted through a number of measurements.

To give some idea of how advanced facial recognition has become. Facebook’s facial recognition research project, known as DeepFace, can identify facial matches in pairs of photographs, without regard to lighting conditions or angle, with 97.25% accuracy. To put this in context, humans can perform the same task with 97.53% accuracy. That’s a difference of only 0.28%.

Speculative in nature, this instance of Anti-Face force feeds fragments of face pictures into a hacked version of Adobe’s photomerge algorithm—an algorithm explicitly built to create panoramas from key marks in landscapes. What it produces are sometimes elegant sometimes monsterous but always multiple. Who is the true subject of the photograph? There is no single identity or owner of this face…

Promotional Process Video

  ANTI–FACE brought to you by Procedural Pleasures Labs  © 2014