WIAMIS 2008 talk: Video summarization through acceleration

With the contribution “Redundancy Removing by Adaptive Acceleration and Event Clustering for Video Summarization” Bernard Mérialdo presented a very interesting approach to video summarization. Beside other methods the video is accelerated. Based on the overall motion the actual amount of acceleration is adapted. Furthermore they apply hierarchical clustering on fixed size segments to select most important parts. Therefore the video is faster when not very much motion is going on and slower if fast motion is characteristics for a video segment. Nice added value was hat he described the RUSHES track of TRECVID and the evaluation process. Experts evaluate the accuracy of the summaries and weight them with extraction time and size of the summary.