| Thesis detail | : |
The objective of this project is to investigate how content-based video retrieval systems may supplement archival metadata and social tagging as search mechanism. The technique for detecting similar clips on video will be explored in order to automatically generate video tagging. Specifically, various content-based video retrieval techniques will be investigated such as, near-duplicate video detection, multimedia event detection, object detection and motion detection.
As this research is part of an ARC Linkage project in conjunction with Circus Oz, proposed content-based retrieval system will be applied to Circus Oz videos. Specifically, a prototype of online Circus Oz video system will be developed. This prototype include video collection server, metadata video database, and web-based application software. |
