R.T.Collins, X.Zhou and S.K.Teh,
"An Open Source Tracking Testbed and Evaluation Web Site,"
IEEE Workshop Performance Evaluation of Tracking and Surveillance (PETS)
Breckenridge, CO, Jan 2005.
We have implemented a GUI-based tracking testbed
system in C and Intel OpenCV, running within the
Microsoft Windows environment. The motivation for the
testbed is to run and log tracking experiments in near
real-time. The testbed allows a tracking experiment to be
repeated from the same starting state using different
tracking algorithms and parameter settings, thereby
facilitating comparison of algorithms. We have also
developed a tracking evaluation web site to encourage
third-party self-evaluation of state-of-the-art tracking
algorithms. The site hosts source code for the tracking
testbed, a set of ground-truth datasets, and a method for
on-line evaluation of uploaded tracking results.
Click here for
full paper (1.1MB, pdf file).