Visual tracking by fusing multiple cues with context-sensitive reliabilities

Erkut Erdem, Séverine Dubuisson, Isabelle Bloch