Videos and ground truth data
----------------------------

You can find here two video sequences which were used to demonstrate the robustness of
our fragments-based tracker. The two sequences are the "face" sequence and the "woman"
sequence.

In the MATLAB .mat file you have the ground truth position in the 
variable sequence_name_gt. The position is the center point of the target, with y-coordinate
in the first row.

The positions were extracted every fifth frame. The corresponding frame numbers are in the
variable sequence_name_frame_nums.

The targets are fixed size with half sizes given by:

woman sequence: h = 33; w = 9;
face sequence: h = 73; w = 58;

Credit
------

We would appreciate it if you cite our work when using these videos and data:

Amit Adam, Ehud Rivlin, Ilan Shimshoni: Robust Fragments-based Tracking using the Integral Histogram.
Proc. CVPR  2006, pp. 798-805