Identifying surprising events in video using bayesian topic models

Avishai Hendel*, Daphna Weinshall, Shmuel Peleg

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

3 Scopus citations

Abstract

In this paper we focus on the problem of identifying interesting parts of the video. To this end we employ the notion of Bayesian surprise, as defined in [9, 10], in which an event is considered surprising if its occurrence leads to a large change in the probability of the world model. We propose to compute this abstract measure of surprise by first modeling a corpus of video events using the Latent Dirichlet Allocation model. Subsequently, we measure the change in the Dirichlet prior of the LDA model as a result of each video event's occurrence. This leads to a closed form expression for an event's level of surprise. We tested our algorithm on a real world video data, taken by a camera observing an urban street intersection. The results demonstrate our ability to detect atypical events, such as a car making a U-turn or a person crossing an intersection diagonally.

Original languageAmerican English
Title of host publicationDetection and Identification of Rare Audiovisual Cues
EditorsDaphna Weinshall, Jorn Anemuller, Luc Gool, Luc Gool
Pages97-105
Number of pages9
DOIs
StatePublished - 2012

Publication series

NameStudies in Computational Intelligence
Volume384
ISSN (Print)1860-949X

Bibliographical note

Funding Information:
Work is funded by the EU Integrated Project DIRAC (IST-027787).

Fingerprint

Dive into the research topics of 'Identifying surprising events in video using bayesian topic models'. Together they form a unique fingerprint.

Cite this