Author | Conference | Journal | Organization | Year | DOI
Look for results that meet for the following criteria:
Academic
Publications
Video summarization via transferrable structured learning
Edit
Video summarization via transferrable structured learning  
BibTex | RIS | RefWorks Download
It is well-known that textual information such as video transcripts and video reviews can significantly enhance the performance of video summarization algorithms. Unfortunately, many videos on the Web such as those from the popular video sharing site YouTube do not have useful textual information. The goal of this paper is to propose a transfer learning framework for video summarization: in the training process both the video features and textual features are exploited to train a summarization algorithm while for summarizing a new video only its video features are utilized. The basic idea is to explore the transferability between videos and their corresponding textual information. Based on the assumption that video features and textual features are highly correlated with each other, we can transfer textual information into knowledge on summarization using video information only. In particular, we formulate the video summarization problem as that of learning a mapping from a set of shots of a video to a subset of the shots using the general framework of SVM-based structured learning. Textual information is transferred by encoding them into a set of constraints used in the structured learning process which tend to provide a more detailed and accurate characterization of the different subsets of shots. Experimental results show significant performance improvement of our approach and demonstrate the utility of textual information for enhancing video summarization.
Conference: World Wide Web Conference Series - WWW , pp. 287-296, 2011
Cumulative Annual
View Publication
The following links allow you to view full publications. These links are maintained by other sources not affiliated with Microsoft Academic Search.
More »
Share this on | Contribute to Academic