KU-ISPL TRECVID 2014 multimedia event detection system

Seongjae Lee, Han Wang, Minseok Keum, Dubok Park, Hyunsik Choi, Zaur Fataliyev, Hanseok Ko

Research output: Contribution to conferencePaperpeer-review

1 Citation (Scopus)

Abstract

In this paper, KU-ISPL's system for TRECVID 2014 Multimedia Event Detection (MED) task is described. The system essentially combines various attributes for scene analysis; audio information for Acoustic Scene Analysis (ASA), Automatic Speech Recognition (ASR), visual information for Visual Scene Analysis (VSA), and Optical Character Recognition (OCR). In the fusion process, heterogeneous data from each module is incorporated, and then the rank of event for each video can be generated according to the final score. In order to verify the effectiveness of our system, we have participated with the following tasks of MED; Pre-specified 010EX, Pre-specified 100EX, Ad-hoc 010EX, and Ad-hoc 100EX. The result of our first participation from NIST shows that the Mean Average Precision (MAP) scores of each MED task were 2.3%, 4.6%, 2.1%, and 2.7%, respectively.

Original languageEnglish
Publication statusPublished - 2020
Event2014 TREC Video Retrieval Evaluation, TRECVID 2014 - Orlando, United States
Duration: 2014 Nov 102014 Nov 12

Conference

Conference2014 TREC Video Retrieval Evaluation, TRECVID 2014
CountryUnited States
CityOrlando
Period14/11/1014/11/12

ASJC Scopus subject areas

  • Information Systems
  • Signal Processing
  • Electrical and Electronic Engineering

Fingerprint Dive into the research topics of 'KU-ISPL TRECVID 2014 multimedia event detection system'. Together they form a unique fingerprint.

Cite this