Human observation and computer analysis of human running gaits.

摘要

There is no perfect gait when it comes to running, however, one would expect that fast, trained runners would share traits in their gaits that untrained runners would lack. Since both people and computers can recognize and distinguish human gaits, we wish to determine what traits humans recognize in trained and untrained runners, and we wish to have a computer recognize the same traits in order to emulate or explain the human observations. By showing models of runners created using motion capture techniques to expert human observers, we learn that humans can't distinguish between trained and untrained runners, but they do notice many different trends among runners. We then use many different computer science techniques to find that K, a dimensionless constant calculated from stride frequency and leg length, as well as Tp, how trained a runner is perceived to be, can both be used to explain the human perceptions.