apennisi / kfct

Kalman Filter Compressive Tracking

Kalman Filter Compressive Tracking

Kalman Filter Compressive Tracking (KFCT) is a modified version of the original Compressive Tracking (CT) presented by Zhang et al. in Real-time Compressive Tracking. In such a version, a Kalman Filter has been implemented for supporting the original CT, where the rectangles, containing the possible candidate, are sampled around the prediction of the CT.

Requirements

• OpenCV

How to build

KCFT works under Linux environments. I recommend a so-called out of source build which can be achieved by the following command sequence:

• mkdir build
• cd build
• cmake ../
• make -j<number-of-cores+1>

How to use

Go to the main KCFT directory and launch one of the scripts:

sh scripts/deer_kalman.sh

Or if you want to use your own set of images, go into the bin folder and launch the script as:

./ct -kalman x y w h dt path/to/the/image/folder

for the Kaman version.

Or:

./ct -origin x y w h path/to/the/image/folder

for the original version.

Where:

• x,y,w,h: are the coordinates of the first rectangle
• dt: is the parameter of the Kalman Filter.