作者: Y. Alp Aslandogan , Clement T. Yu
关键词: Computer science 、 Information retrieval 、 Set (abstract data type) 、 World Wide Web 、 Image retrieval 、 Image (mathematics) 、 Identification (information) 、 Context (language use) 、 Face detection 、 Web search engine 、 Search engine
摘要: In this work, we examine evidence combination mechanisms for classifying multimedia information. particular, linear and Dempster-Shafer methods of in the context identifying personal images on World Wide Web. An automatic web search engine named Diogenes1 searches combines different pieces identification. The sources consist input from face detection/recognition text/HTML analysis modules. A degree uncertainty is involved with both these sources. Diogenes automatically determines locally each retrieval uses information to set a relative significance evidence. To our knowledge, first image using based object recognition dynamic local assessment. experiments comfortably outperformed some well known commercial research prototype engines celebrity queries.