Video summarization at Brno university of technology

This paper describes the video summarization system built for the TRECVID 2008 evaluation by the Brno team. Motivations for the system design and its overall structure are described followed by more detailed description of the critical parts of the system. Low-level features, which are extracted from each frame, are clustered to group visually similar shots together. The final video summary production is an iterative procedure, where the probability, speed and trimming of each cluster candidate are evaluated until some criteria, such as final summary length, are fulfilled. The paper also contains the discussion about appropriate layout of the final video summary, taking into account experiences from the last TRECVID evaluation. The final conclusion points out the weak and strong aspects of the presented approach reflecting system performance in comparison with other state-of-the-art systems.