An attribute based news video indexing

Ichiro Ide, Reiko Hamada, Shuichi Sakai, Hideohiko Tanaka
2001 Proceedings of the 2001 ACM workshops on Multimedia multimedia information retrieval - MULTIMEDIA '01  
We propose an attribute based news video indexing method. Semantic attributes of both text and image are analyzed, and correspondences within each attribute is considered when indexing, to ensure correspondences between indices and image contents. In this paper, both text and image attribute analysis is briefly introduced, and the result of the final indexing is displayed. The overall result showed comparable performance to existing systems in the case of personal name -character indexing.
doi:10.1145/500933.500955 dblp:conf/mm/IdeHST01 fatcat:i2uptdm6cjccnmjkt2v7ilmgy4