A dynamic probabilistic multimedia retrieval model

Author

Ianeva, Tzvetanka I. ; De Vries, Arjen P. ; Westerveld, Thijs

Author_Institution

Departamento de Informatica, Valencia Univ., Spain

Volume

3

fYear

2004

fDate

27-30 June 2004

Firstpage

1607

Abstract

We describe the application of a probabilistic multimedia model to video retrieval. From video shots, we compute Gaussian-mixture models that capture correlations in time and space, such as the appearance and disappearance of objects. These models improve the precision of "query by example/s" results in the TRECVID 2003 collection when compared to models that only make use of static visual information. Furthermore, integrated with information from automatic speech recognition (ASR) transcripts, they outperform ASR only results.

Keywords

Gaussian distribution; correlation methods; information retrieval; speech recognition; video databases; video signal processing; ASR transcripts; Gaussian-mixture models; automatic speech recognition; dynamic probabilistic multimedia retrieval model; object appearance/disappearance; query by example; spatio-temporal video retrieval model; time/space correlations; video search systems; Automatic speech recognition; Data mining; Gaussian processes; Humans; Image color analysis; Image retrieval; Information retrieval; Layout; Motion analysis; NIST;

fLanguage

English

Publisher

ieee

Conference_Titel

Multimedia and Expo, 2004. ICME '04. 2004 IEEE International Conference on

Print_ISBN

0-7803-8603-5

Type

conf

DOI

10.1109/ICME.2004.1394557

Filename

1394557