National Institute of Informatics, Japan at TRECVID 2011

This paper reports our experiments for three TRECVID 2011 tasks: instance search, semantic indexing, and multimedia event detection. For the instance search task, we present three different approaches: (i) Large vocabulary quantiza- tion by hierarchical k-means and weighted histogram intersection based ranking metric (ii) Combination of similarities based on Glocal quantization of two set of SIFTs and color histograms from the full frames, and (iii) Keypoint matching is used to compute the similarity between images of the query and images of all videos. For the semantic indexing task and the multimedia event detection task, we report the experiments using NII-KAORI- SECODE framework. Our approaches can be considered as one of the baseline approaches for evaluation of these tasks.