IBM research TRECVID-2005 video retrieval system

Abstract

In this paper, we describe the IBM Research system for indexing, analysis, and retrieval of video as applied to the TREC-2005 video retrieval benchmark. We participated in the shot boundary detection, high-level feature extraction and search tasks and performed several new experiments in all the tasks. The paper describes the details of the approaches as well as the performance analysis. In general we observed good performance across all three tasks. In the detection task, we were able to achieve top mean average precision performance for all over 7 systems. In automatic search, we were able to achieve top mean average precision performance for 4 of our 6 automatic runs.