How features of the Human Face Affect Recognition
Ross Beveridge, Computer Science, CSU

What factors make some human subjects easier to recognize than others?
Just as important, how does one set about trying to answer such a
question? This talk will review work here at CSU on using statistical
inference methods to better understanding human identification systems.
I will describe our work developing the CSU Face Identification
Evaluation System. I will also describe a series of experiments aimed at
better understanding how common face recognition algorithms behave. The
most recent experiment addresses human subject covariates, such things
as whether a subject is old or young, male or female, etc., and how
these affect the performance of three common face recognition algorithms.