Affective classification of movie scenes based on two-pass clustering technique

Che Wei Sung, Yu Chieh Wu, Kun Huang Chien, Jie Chi Yang

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

Abstract

Video-based learning not only provides rich content but also gives multimedia e-learning environment. In this paper, we present an affective movie classification tool, which automatically segments and labels emotion tags for the given video film. Our method integrates nine audio and visual features from each input video. Then the proposed two-pass clustering technique is used to group similar video scenes and gives labels. One good property of our method is that the need of manual annotated training data is un-required. We compared with the other famous algorithms such as ART2 and K-means. The experimental result shows that our video affect classification tool yields better accuracy (recall and precision) than the other clustering approaches. In short, it achieves ∼80% in F-measure rate for 119 testing scenes.

Original languageEnglish
Title of host publicationWorkshop Proceedings of the 17th International Conference on Computers in Education, ICCE 2009
Pages60-64
Number of pages5
StatePublished - 2009
Event17th International Conference on Computers in Education, ICCE 2009 - Hong Kong, Hong Kong
Duration: 30 Nov 20094 Dec 2009

Publication series

NameWorkshop Proceedings of the 17th International Conference on Computers in Education, ICCE 2009

Conference

Conference17th International Conference on Computers in Education, ICCE 2009
Country/TerritoryHong Kong
CityHong Kong
Period30/11/094/12/09

Keywords

  • Self-organizing feature map
  • Video content-based analysis

Fingerprint

Dive into the research topics of 'Affective classification of movie scenes based on two-pass clustering technique'. Together they form a unique fingerprint.

Cite this