National Institute of Informatics, Japan at TRECVID 2009

This paper reports our experiments for TRECVID 2009 tasks: high level feature extraction, search and contentbased copy detection. For the high level feature extraction task, we used the baseline features such as color moments, edge orientation histogram, local binary patterns and local features trained with SVM classifiers and nearest neighbor classifiers. For the search task, we used . Concerning content based video copy detection (CBVCD), using local features leads to good robustness to most types of photometric or geometric transformations. However, to achieve both good precision and good recall when the transformations are strong, especially occlusions, feature configurations should be taken into account. This usually leads to complex matching operations that are incompatible with scalable copy detection. We suggest a computationally inexpensive solution for including a minimal amount of configuration information that significantly improves the balance between overall detection quality and scalability. I. HIGH LEVEL FEATURE EXTRACTION