Authors: Andres Romero y Tieran, Michèle Gouiffès and Lionel Lacassagne

The tracking is initialized by FoT. Then CT is used when necessary: when the target is lost or the target has a non-rigid motion. In the result sequences, the bounding box around the target is red when FoT is used and blue when CT is used. Examples of sequences (the dataset comes from http://info.ee.surrey.ac.uk/Personal/Z.Kalal/tld.html)

• 06_car
  


<object style="width:480px;height:390px"> <param name="movie" value="video/player.swf?file=auto.flv" /> <param name="quality" value="high" /> </object>
• 07_motocross
  


<object style="width:480px;height:390px"> <param name="movie" value="video/player.swf?file=moto.flv" /> <param name="quality" value="high" /> </object>
• 03_pedestrian1
  


<object style="width:480px;height:390px"> <param name="movie" value="video/player.swf?file=pedestrian1.flv" /> <param name="quality" value="high" /> </object>
• 10_panda
  


<object style="width:480px;height:390px"> <param name="movie" value="video/player.swf?file=panda.flv" /> <param name="quality" value="high" /> </object>