Activity Retrieval in Large Surveillance Videos*

Greg Castanon*, Pierre-Marc Jodoin, Venkatesh Saligrama* and Andre Caron,    *Boston University Boston, MA, USA,    Université de Sherbrooke, Sherbrooke, Canada

Abstract

We present a fast and flexible content-based retrieval method for surveillance video. Designing a video search robust to uncertain activity duration, high variability in object shapes and scene content is challenging. We propose a four-step approach to video search: (1) light-weight pre-processing to extract local video features, (2) inverted indexing scheme based on locality-sensitive hashing (LSH) to accelerate retrieval, (3) a query interface to convert keywords into feature-based queries and (4) novel dynamic programming ...

Get Academic Press Library in Signal Processing now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.