Detecting events by clustering videos from large media databases

Authors:
Jarmo Makkonen;Riitta Kerminen;Igor D.D. Curcio;Sujeet Mate;Ari Visa
Affiliations:
Tampere University of Technology, Tampere, Finland;Tampere University of Technology, Tampere, Finland;Nokia Research Center, Tampere, Finland;Nokia Research Center, Tampere, Finland;Tampere University of Technology, Tampere, Finland
Venue:
Proceedings of the 2nd ACM international workshop on Events in multimedia
Year:
2010

Citing 6
Cited 2

World explorer: visualizing aggregate data from unstructured text in geo-referenced collections

Proceedings of the 7th ACM/IEEE-CS joint conference on Digital libraries
Towards automatic extraction of event and place semantics from flickr tags

SIGIR '07 Proceedings of the 30th annual international ACM SIGIR conference on Research and development in information retrieval
How flickr helps us make sense of the world: context and content in community-contributed media collections

Proceedings of the 15th international conference on Multimedia
Generating diverse and representative image search results for landmarks

Proceedings of the 17th international conference on World Wide Web
Introduction to Information Retrieval

Introduction to Information Retrieval
Learning similarity metrics for event identification in social media

Proceedings of the third ACM international conference on Web search and data mining

Effective web video clustering using playlist information

Proceedings of the 27th Annual ACM Symposium on Applied Computing
E-LAMP: integration of innovative ideas for multimedia event detection

Machine Vision and Applications

Quantified Score

Hi-index	0.00

Visualization

Abstract

As the number of user-generated videos rises in the Internet, there is a growing need for more efficient search tools that enable the users to find the desired content. Moreover, the associated video metadata for the content is often incomplete or even misleading. This paper addresses the problem of finding events by utilizing the video metadata from a video database by proposing two novel methods that are used in parallel. The first one is missing data compensation, which harvests missing data values from the textual descriptions in the video metadata. The second one is a layered clustering method that divides the videos in the database into clusters, each of which is considered as an event. The methods are tested with manually selected data from YouTube. The results show that missing data compensation yields better results in terms of accuracy than using ram data, and that the clustering method provides acceptable results and is a promising approach for further research.