Real-time Spatio-Temporal Action Localization in 360 Videos

Bo Chen, Ahmed Ali-Eldin, Prashant Shenoy, Klara Nahrstedt

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Spatio-temporal action localization of human actions in a video has been a popular topic over the past few years. It tries to localize the bounding boxes, the time span and the class of one action, which summarizes information in the video and helps humans understand it. Though many approaches have been proposed to solve this problem, these efforts have only focused on perspective videos. Unfortunately, perspective videos only cover a small field-of-view (FOV), which limits the capability of action localization. In this paper, we develop a comprehensive approach to real-time spatio-temporallocalization that can be used to detect actions in 360 videos. We create two datasets named UCF-101-24-360 and JHMDB-21-360 for our evaluation. Our experiments show that our method consistently outperforms other competing approaches and achieves a real-time processing speed of 15fps for 360 videos.

Original languageEnglish (US)
Title of host publicationProceedings - 2020 IEEE International Symposium on Multimedia, ISM 2020
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages73-76
Number of pages4
ISBN (Electronic)9781728186979
DOIs
StatePublished - Dec 2020
Event22nd IEEE International Symposium on Multimedia, ISM 2020 - Virtual, Naples, Italy
Duration: Dec 2 2020Dec 4 2020

Publication series

NameProceedings - 2020 IEEE International Symposium on Multimedia, ISM 2020

Conference

Conference22nd IEEE International Symposium on Multimedia, ISM 2020
Country/TerritoryItaly
CityVirtual, Naples
Period12/2/2012/4/20

Keywords

  • 360 video
  • Spatio temporal action localization

ASJC Scopus subject areas

  • Media Technology

Fingerprint

Dive into the research topics of 'Real-time Spatio-Temporal Action Localization in 360 Videos'. Together they form a unique fingerprint.

Cite this