Affordable Access

Access to the full text

Interactive video summarization with human intentions

Authors
  • Liu, Huaping1
  • Sun, Fuchun1
  • Zhang, Xinyu2
  • Fang, Bin1
  • 1 Tsinghua University, BNRist, State Key Lab. of Intelligent Technology and Systems, Department of Computer Science and Technology, Beijing, China , Beijing (China)
  • 2 Tsinghua University, State Key Laboratory of Automotive Safety and Energy, Beijing, China , Beijing (China)
Type
Published Article
Journal
Multimedia Tools and Applications
Publisher
Springer-Verlag
Publication Date
Jun 30, 2018
Volume
78
Issue
2
Pages
1737–1755
Identifiers
DOI: 10.1007/s11042-018-6305-x
Source
Springer Nature
Keywords
License
Yellow

Abstract

Automatic video summarization, which is a typical cognitive-inspired task and attempts to select a small set of the most representative images or video clips for a specific video sequence, is therefore vital for enabling many tasks. In this work, we develop an interactive Non-negative Matrix Factorization (NMF) method for representative action video discovery. The original video is first evenly segmented into short clips, and the bag-of-words model is used to describe each clip. A temporally consistent NMF model is subsequently used for clustering and action segmentation. Because the clustering and segmentation results may not satisfy user intention, the user-controlled operations MERGE and ADD are developed to permit the user to adjust the results in line with expectations. The newly developed interactive NMF method can therefore generate personalized results.Experimental results on the public Weizman dataset demonstrate that our approach provides satisfactory action discovery and segmentation results.

Report this publication

Statistics

Seen <100 times