Extending a Distributed Online Machine Learning Framework for Streaming Video Analysis

With the advent of video processing and Internet technologies, tremendous number of video clips are now accessible on the Internet. However, due to large video sizes and real-time video streaming, it is very difficult to analyze video in real time, which is a necessary step in many video web services. In this paper, we propose a novel extension method for an existing distributed machine learning framework, Jubatus, which was mainly developed for analyzing textual data. With our extension, numerous video clips can be analyzed efficiently by a machine learning framework. We also describe some preliminary evaluation results to indicate the efficiency of our proposed extension.