TY - GEN

T1 - Finding pictures of objects in large collections of images

AU - Forsyth, David A.

AU - Malik, Jitendra

AU - Fleck, Margaret M.

AU - Greenspan, Hayit

AU - Leung, Thomas

AU - Belongie, Serge

AU - Carson, Chad

AU - Bregler, Chris

N1 - Funding Information: We would like to thank R. Blasi and K. Murphy who collaborated with S. Belongie in the work on learning decision trees for visual concept classification. We thank Joe Mundy for suggesting that the response of a grouper may indicate the presence of an object. Aspects of this research were supported by the National Science Foundation under grants IRI-9209728, IRI-9420716, IRI-9501493, an NSF Young Investigator award, an NSF Digital Library award IRI-9411334, an instrumentation award CDA-9121985, and by a Berkeley Fellowship. Publisher Copyright: © 1996, Springer Verlag. All rights reserved.

PY - 1996

Y1 - 1996

N2 - Retrieving images from very large collections, using image content as a key, is becoming an important problem. Users prefer to ask for pictures using notions of content that are strongly oriented to the presence of abstractly defined objects. Computer programs that implement these queries automatically are desirable, but are hard to build because conventional object recognition techniques from computer vision cannot recognize very general objects in very general contexts. This paper describes our approach to object recognition, which is structured around a sequence of increasingly specialized grouping activities that assemble coherent regions of image that can be shown to satisfy increasingly stringent constraints. The constraints that are satisfied provide a form of object classification in quite general contexts. This view of recognition is distinguished by: far richer involvement of early visual primitives, including color and texture; hierarchical grouping and learning strategies in the classification process; the ability to deal with rather general objects in uncontrolled configurations and contexts. We illustrate these properties with four case-studies: one demonstrating the use of color and texture descriptors; one showing how trees can be described by fusing texture and geometric properties; one learning scenery concepts using grouped features; and one showing how this view of recognition yields a program that can tell, quite accurately, whether a picture contains naked people or not.

AB - Retrieving images from very large collections, using image content as a key, is becoming an important problem. Users prefer to ask for pictures using notions of content that are strongly oriented to the presence of abstractly defined objects. Computer programs that implement these queries automatically are desirable, but are hard to build because conventional object recognition techniques from computer vision cannot recognize very general objects in very general contexts. This paper describes our approach to object recognition, which is structured around a sequence of increasingly specialized grouping activities that assemble coherent regions of image that can be shown to satisfy increasingly stringent constraints. The constraints that are satisfied provide a form of object classification in quite general contexts. This view of recognition is distinguished by: far richer involvement of early visual primitives, including color and texture; hierarchical grouping and learning strategies in the classification process; the ability to deal with rather general objects in uncontrolled configurations and contexts. We illustrate these properties with four case-studies: one demonstrating the use of color and texture descriptors; one showing how trees can be described by fusing texture and geometric properties; one learning scenery concepts using grouped features; and one showing how this view of recognition yields a program that can tell, quite accurately, whether a picture contains naked people or not.

UR - http://www.scopus.com/inward/record.url?scp=84979074103&partnerID=8YFLogxK

U2 - 10.1007/3-540-61750-7_36

DO - 10.1007/3-540-61750-7_36

M3 - Conference article

AN - SCOPUS:84979074103

SP - 335

EP - 360

JO - Lecture Notes in Computer Science

JF - Lecture Notes in Computer Science

SN - 0302-9743

T2 - International Workshop on Object Representation in Computer Vision II, ECCV 1996

Y2 - 13 April 1996 through 14 April 1996

ER -