IVIST: Interactive Video Search Tool in VBS 2022

Sangmin Lee, Sungjune Park, Yong Man Ro

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

1 Scopus citations

Abstract

This paper presents the details of the proposed video retrieval tool, named Interactive VIdeo Search Tool (IVIST) for the Video Browser Showdown (VBS) 2022. In order to retrieve desired videos from a multimedia database, it is necessary to match queries from humans and video shots in the database effectively. To boost such matching relationship, we propose a multi-modal-based retrieval scheme that can fully utilize various modal features of the multimedia data and synthetically consider the matching relationships between modalities. The proposed IVIST maps human-made queries (e.g., language) and features (e.g., visual and sound) from the database into a multi-modal matching latent space through deep neural networks. Based on the latent space, videos with high similarity to the query feature are suggested as candidate shots. Prior knowledge-based filtering can be further applied to refine the results of candidate shots. Moreover, the user interface of the tool is devised in a user-friendly way for interactive video searching.

Original languageEnglish
Title of host publicationMultiMedia Modeling - 28th International Conference, MMM 2022, Proceedings
EditorsBjörn Þór Jónsson, Cathal Gurrin, Minh-Triet Tran, Duc-Tien Dang-Nguyen, Anita Min-Chun Hu, Binh Huynh Thi Thanh, Benoit Huet
PublisherSpringer Science and Business Media Deutschland GmbH
Pages524-529
Number of pages6
ISBN (Print)9783030983543
DOIs
StatePublished - 2022
Externally publishedYes
Event28th International Conference on MultiMedia Modeling, MMM 2022 - Phu Quoc, Viet Nam
Duration: 6 Jun 202210 Jun 2022

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume13142 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference28th International Conference on MultiMedia Modeling, MMM 2022
Country/TerritoryViet Nam
CityPhu Quoc
Period6/06/2210/06/22

Keywords

  • Interactive video retrieval
  • Multi-modal matching
  • Video Browser Showdown

Fingerprint

Dive into the research topics of 'IVIST: Interactive Video Search Tool in VBS 2022'. Together they form a unique fingerprint.

Cite this