| Paper: | MMSP-P3.1 | ||
| Session: | Multimedia Database, Content Retrieval, Joint Processing and Standards | ||
| Time: | Wednesday, May 17, 16:30 - 18:30 | ||
| Presentation: | Poster | ||
| Topic: | Multimedia Signal Processing: Content-based information retrieval and pattern discovery | ||
| Title: | An Automatic Video Semantic Annotation Scheme Based on Combination of Complementary Predictors | ||
| Authors: | Yan Song, University of Science and Technology of China, China; Xian-Sheng Hua, Microsft Research Asia, China; Li-Rong Dai, Meng Wang, Ren-Hua Wang, University of Science and Technology of China, China | ||
| Abstract: | Given a large set of video database, to connect video segments with a certain set of semantic concepts with least manual labors is an elementary step for video indexing and searching. However, due to the large gap between high-level semantics and low-level features, it is difficult to obtain high accuracy annotation automatically. In this paper, we propose a novel automatic video annotation framework, which improves the annotation performance by learning from unlabeled samples and exploring local consistency and temporal relationship of video sequences. To effectively learn from unlabeled data, a sample selection scheme based on combining complementary predictors is proposed, which iteratively refines the performance of the initial predictors in the learning process. And a filtering-based method is applied to further improve the annotation accuracy, in which video temporal consistency is sufficiently exploited. Experiment results show that the proposed automatic video annotation method performs superior to the general learning-based method and the typical co-training method. | ||
©2018 Conference Management Services, Inc. -||- email: webmaster@icassp2006.org -||- Last updated Friday, August 17, 2012