Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

Region based image annotation through multiple-instance learning

Region based image annotation through multiple-instance learning Region Based Image Annotation Through Multiple-Instance Learning Changbo Yang, Ming Dong and Farshad Fotouhi Department of Computer Science Wayne State University Detroit, MI 48202 cbyang, mdong, fotouhi@cs.wayne.edu ABSTRACT In an annotated image database, keywords are usually associated with images instead of individual regions, which poses a major challenge for any region based image annotation algorithm. In this paper, we propose to learn the correspondence between image regions and keywords through Multiple-Instance Learning (MIL). After a representative image region has been learned for a given keyword, we consider image annotation as a problem of image classi cation, in which each keyword is treated as a distinct class label. The classi cation problem is then addressed using the Bayesian framework. The proposed image annotation method is evaluated on an image database with 5, 000 images. Categories and Subject Descriptors I.4.8 [Image Processing and Computer Vision]: Scene Analysis-object recognition, H.2.8 [Database Management]: Database Applications - image databases. General Terms: Algorithms, Measurement, Experimentation Keywords: Automatic image annotation, Multiple-Instance Learning Figure 1: Three sample images of œtiger  (left column) and their segmented regions (right column). A large number of irrelevant noisy regions, such as œgrass , œwater , and œbush , exist http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png

Region based image annotation through multiple-instance learning

Association for Computing Machinery — Nov 6, 2005

Loading next page...
/lp/association-for-computing-machinery/region-based-image-annotation-through-multiple-instance-learning-UWhu073VJO

References (12)

Datasource
Association for Computing Machinery
Copyright
Copyright © 2005 by ACM Inc.
ISBN
1-59593-044-2
doi
10.1145/1101149.1101245
Publisher site
See Article on Publisher Site

Abstract

Region Based Image Annotation Through Multiple-Instance Learning Changbo Yang, Ming Dong and Farshad Fotouhi Department of Computer Science Wayne State University Detroit, MI 48202 cbyang, mdong, fotouhi@cs.wayne.edu ABSTRACT In an annotated image database, keywords are usually associated with images instead of individual regions, which poses a major challenge for any region based image annotation algorithm. In this paper, we propose to learn the correspondence between image regions and keywords through Multiple-Instance Learning (MIL). After a representative image region has been learned for a given keyword, we consider image annotation as a problem of image classi cation, in which each keyword is treated as a distinct class label. The classi cation problem is then addressed using the Bayesian framework. The proposed image annotation method is evaluated on an image database with 5, 000 images. Categories and Subject Descriptors I.4.8 [Image Processing and Computer Vision]: Scene Analysis-object recognition, H.2.8 [Database Management]: Database Applications - image databases. General Terms: Algorithms, Measurement, Experimentation Keywords: Automatic image annotation, Multiple-Instance Learning Figure 1: Three sample images of œtiger  (left column) and their segmented regions (right column). A large number of irrelevant noisy regions, such as œgrass , œwater , and œbush , exist

There are no references for this article.