Title
University Of Central Florida At Trecvid 2006 High-Level Feature Extraction And Video Search
Abstract
In this paper, we describe our experiments in high-level features extraction and interactive topic search tasks of TRECVID 2006. We designed a unified high-level features extraction framework for the 39 high-level features. Various low-level visual features were extracted from the key-frames of the shots. Then the SVM classifiers were trained fore. The final classification results were produced by fusing and combining these classifiers. The experiment results show that the combined classifiers substantially improved the performance over the individual feature based classifier. In topic search task, we improved our PEGASUS news video retrieval system, which has friendly user interface, fast indexing and various relevance feedback mechanisms. Based on the evaluation results, this year's topic search results are better compared to last year.
Publication Date
1-1-2006
Publication Title
2006 TREC Video Retrieval Evaluation Notebook Papers
Number of Pages
-
Document Type
Article; Proceedings Paper
Personal Identifier
scopus
Copyright Status
Unknown
Socpus ID
84905178815 (Scopus)
Source API URL
https://api.elsevier.com/content/abstract/scopus_id/84905178815
STARS Citation
Liu, Jingen; Zhai, Yun; Basharat, Arslan; Orhan, Bilal; and Khan, Saad M., "University Of Central Florida At Trecvid 2006 High-Level Feature Extraction And Video Search" (2006). Scopus Export 2000s. 8950.
https://stars.library.ucf.edu/scopus2000/8950