Unsupervised TV Program Structuring
Résumé rédigé par
Directeur de thèse:
Unité de recherche UMR 7102 Laboratoire de recherche d'EURECOM
An important feature for automatic segmentation and structuring of TV programs is to provide tools that allow users to efficiently navigate within TV programs, that is, directly seek a certain part of a program or particular moments of interest. To achieve that, the original structure of each TV program needs to be recovered. In other terms, the start and end times of each part of the program have to be detected. This processing step cannot be performed manually when dealing with the huge amount of available TV programs. This thesis addresses the problem of TV program structuring and proposes to provide a tool that structures in a completely unsupervised way a wide range of TV programs.