Skip to main navigation Skip to search Skip to main content

3D convolutional network based foreground feature fusion

  • Xi'an Jiaotong University

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

3 Scopus citations

Abstract

With explosion of videos, action recognition has become an important research subject. This paper makes a special effort to investigate and study 3D Convolutional Network. Focused on the problem of ConvNet dependence on multiple large scale dataset, we propose a 3D ConvNet structure which incorporate the original 3D-ConvNet features and foreground 3D-ConvNet features fused by static object and motion detection. Our architecture is trained and evaluated on the standard video actions benchmarks of UCF-101 and HMDB-51, experimental results demonstrate that with merely 50% pixels utilization, foreground ConvNet achieves satisfying performance as same as origin. With feature fusion, we achieve 83.7% accuracy on UCF-101 exceeding original ConvNet.

Original languageEnglish
Title of host publicationProceedings - 2018 IEEE International Symposium on Multimedia, ISM 2018
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages253-258
Number of pages6
ISBN (Electronic)9781538668573
DOIs
StatePublished - 2 Jul 2018
Event20th IEEE International Symposium on Multimedia, ISM 2018 - Taichung, Taiwan, Province of China
Duration: 10 Dec 201812 Dec 2018

Publication series

NameProceedings - 2018 IEEE International Symposium on Multimedia, ISM 2018

Conference

Conference20th IEEE International Symposium on Multimedia, ISM 2018
Country/TerritoryTaiwan, Province of China
CityTaichung
Period10/12/1812/12/18

Keywords

  • C3D
  • Feature Fusion
  • Foreground Exaction

Fingerprint

Dive into the research topics of '3D convolutional network based foreground feature fusion'. Together they form a unique fingerprint.

Cite this