Face Recognition Engine

XiD Tech won a competition for all Asian and Australian compunies in security market.

 XiD has this powerful technology to create multiple (several hundreds) possible appearences of a person based on one or just a few facial images.  This is done using their proprietary technology that combines computer vision and computer graphics.  To recogniza a person then the system could use this set of potential images of the person’s face which may include different environmental conditions but could also include glasses, facial hair, and other similar changes.

I believe such techn ology could be core to many systems that are likely to emerge not only in security but in many other applications.  What could be more important than a human face to us, humans, in our applications?  I find it surprising that similar ideas have not been used much.  This is an oethognal idea to most other recognition approaches in which the system relies on many faces of the person and then tries to use statistical learning techniques to identify parameters that will help in recognition.  But that approach is debilitating because in most cases, such data may not be avaiolable.  Also, when a completely new situation is encountered, data will not be available.  In Xid like approaches, based on synthesis rather than only learning, one could create scenarios and hence also create the data.

(Disclaimer: I am an advisor to XiD.)



Leave a Reply