A multiple instance approach for keyword-based retrieval in un-annotated image database

Authors:
Jun Jiao;Chen Shen;Bo Dai;Xuan Mo
Affiliations:
Computer Science & Technology Department, Nanjing University, Nanjing, China;School of Software and Electronics, Peking University, Beijing, China;Institute of Automation, Chinese Academy of Sciences, Beijing, China;Department of Automation, Tsinghua University, Beijing, China
Venue:
MMM'10 Proceedings of the 16th international conference on Advances in Multimedia Modeling
Year:
2010

Citing 4
Cited 0

Blobworld: Image Segmentation Using Expectation-Maximization and Its Application to Image Querying

IEEE Transactions on Pattern Analysis and Machine Intelligence
Web image learning for searching semantic concepts in image databases

Proceedings of the 13th international World Wide Web conference on Alternate track papers & posters
Estimating the Support of a High-Dimensional Distribution

Neural Computation
Exploiting image contents in web search

IJCAI'07 Proceedings of the 20th international joint conference on Artifical intelligence

Quantified Score

Hi-index	0.00

Visualization

Abstract

In image retrieval, if user can describe their query concepts by keywords, search results can be returned efficiently and precisely by matching query keywords with text annotation in image databases. However, even if the query keyword is given, keyword-based retrieval can not be applied directly in an image database without any text annotation. The development of Web mining and searching techniques has enabled us to search images in Web by keywords. Thus, we can search the query keywords given by user through Web to obtain example images, and then find those images relevant to user's query in image database with the help of these example images. In order to improve the image retrieval performance, we adopt multiple instance learning when calculating the similarity between example images and images in database. Experiments validate that our method can effectively improve the retrieval performance in un-annotated image database.