A Unified Framework for Event Summarization and Rare Event Detection

A novel approach for event summarization and rare event detection is proposed. Unlike conventional methods that deal with event summarization and rare event detection independently, we solve them together by transforming the problems into a graph editing framework. In our approach, a video is represented as a graph, in which each node of the graph indicates an event obtained by segmenting the video spatially and temporally, while edges between nodes describe the events related to each other. Based on the degree of relations, edges have different weights. After learning the graph structure, our method edits the graph by merging its subgraphs or pruning its edges. The graph is edited toward minimizing a predefined energy model with the Data-Driven Markov Chain Monte Carlo method. The energy model consists of several parameters that represent causality, frequency, and significance of events. We design a specific energy model utilizing these parameters to satisfy each objective of event ...

• Type: Conference
• Year: 2012
• Where: CVPR
• Authors: Junseok Kwon and Kyoung Mu Lee
• Keywords: Markov Chain Monte Carlo, Rare Event Detection, Video Summarization