Via mashable.com/2012/03/23/hitachi-face-recognition/
Local Feature Analysis (LFA) and the speed of nodal point databases has always been very fast.
Now from the camera in real time.
Via mashable.com/2012/03/23/hitachi-face-recognition/
Local Feature Analysis (LFA) and the speed of nodal point databases has always been very fast.
Now from the camera in real time.
Via: wired.com/gadgetlab/2011/04/cops-cyborg-shades/
The “RoboCop” glasses use facial recognition software to spot people of interest in a crowd.
400 faces per second, up to 50 yards away, using 46,000 biometric points?
Now how fast could a set of say 14 to 22 nodal points be?
Found a random company with a few numbers and some Local Feature Analysis stats
http://www.bayometric.com/products/Face-Recognition-SDK.htm
‘Up to 60 million per minute CPU for vector template … using all 3 templates, aggregate search speed up to 10 million per minute per CPU, depending on hardware”