Dynamic learning from multiple examples for semantic object segmentation and search. Xu, Y., Saber, E, & Tekalp, A. COMPUTER VISION AND IMAGE UNDERSTANDING, 95(3):334-353, SEP, 2004. doi abstract bibtex We present a novel ``dynamic learning'' approach for an intelligent image database system to automatically improve object segmentation and labeling without user intervention, as new examples become available, for object-based indexing. The proposed approach is an extension of our earlier work on ``learning by example,'' which addressed labeling of similar objects in a set of database images based on a single example. The proposed dynamic learning procedure utilizes multiple example object templates to improve the accuracy of existing object segmentations and labels. Multiple example templates may be images of the same object from different viewing angles, or images of related objects. This paper also introduces a new shape similarity metric called normalized area of symmetric differences (NASD), which has desired properties for use in the proposed ``dynamic learning'' scheme, and is more robust against boundary noise that results from automatic image segmentation. Performance of the dynamic learning procedures has been demonstrated by experimental results. (C) 2004 Elsevier Inc. All rights reserved.
@article{ ISI:000223379500004,
Author = {Xu, YW and Saber, E and Tekalp, AM},
Title = {{Dynamic learning from multiple examples for semantic object segmentation
and search}},
Journal = {{COMPUTER VISION AND IMAGE UNDERSTANDING}},
Year = {{2004}},
Volume = {{95}},
Number = {{3}},
Pages = {{334-353}},
Month = {{SEP}},
Abstract = {{We present a novel ``dynamic learning{''} approach for an intelligent
image database system to automatically improve object segmentation and
labeling without user intervention, as new examples become available,
for object-based indexing. The proposed approach is an extension of our
earlier work on ``learning by example,{''} which addressed labeling of
similar objects in a set of database images based on a single example.
The proposed dynamic learning procedure utilizes multiple example object
templates to improve the accuracy of existing object segmentations and
labels. Multiple example templates may be images of the same object from
different viewing angles, or images of related objects. This paper also
introduces a new shape similarity metric called normalized area of
symmetric differences (NASD), which has desired properties for use in
the proposed ``dynamic learning{''} scheme, and is more robust against
boundary noise that results from automatic image segmentation.
Performance of the dynamic learning procedures has been demonstrated by
experimental results. (C) 2004 Elsevier Inc. All rights reserved.}},
DOI = {{10.1016/j.cviu.2004.04.003}},
ISSN = {{1077-3142}},
Unique-ID = {{ISI:000223379500004}},
}
Downloads: 0
{"_id":"KumFbtEtE2mkaL6zK","bibbaseid":"xu-saber-tekalp-dynamiclearningfrommultipleexamplesforsemanticobjectsegmentationandsearch-2004","downloads":0,"creationDate":"2016-12-09T13:41:59.969Z","title":"Dynamic learning from multiple examples for semantic object segmentation and search","author_short":["Xu, Y.","Saber, E","Tekalp, A."],"year":2004,"bibtype":"article","biburl":"https://drive.google.com/uc?export=download&id=1d5Wvl98W_buJq6prXBz16a_6uF9Cdt-7","bibdata":{"bibtype":"article","type":"article","author":[{"propositions":[],"lastnames":["Xu"],"firstnames":["YW"],"suffixes":[]},{"propositions":[],"lastnames":["Saber"],"firstnames":["E"],"suffixes":[]},{"propositions":[],"lastnames":["Tekalp"],"firstnames":["AM"],"suffixes":[]}],"title":"Dynamic learning from multiple examples for semantic object segmentation and search","journal":"COMPUTER VISION AND IMAGE UNDERSTANDING","year":"2004","volume":"95","number":"3","pages":"334-353","month":"SEP","abstract":"We present a novel ``dynamic learning'' approach for an intelligent image database system to automatically improve object segmentation and labeling without user intervention, as new examples become available, for object-based indexing. The proposed approach is an extension of our earlier work on ``learning by example,'' which addressed labeling of similar objects in a set of database images based on a single example. The proposed dynamic learning procedure utilizes multiple example object templates to improve the accuracy of existing object segmentations and labels. Multiple example templates may be images of the same object from different viewing angles, or images of related objects. This paper also introduces a new shape similarity metric called normalized area of symmetric differences (NASD), which has desired properties for use in the proposed ``dynamic learning'' scheme, and is more robust against boundary noise that results from automatic image segmentation. Performance of the dynamic learning procedures has been demonstrated by experimental results. (C) 2004 Elsevier Inc. All rights reserved.","doi":"10.1016/j.cviu.2004.04.003","issn":"1077-3142","unique-id":"ISI:000223379500004","bibtex":"@article{ ISI:000223379500004,\nAuthor = {Xu, YW and Saber, E and Tekalp, AM},\nTitle = {{Dynamic learning from multiple examples for semantic object segmentation\n and search}},\nJournal = {{COMPUTER VISION AND IMAGE UNDERSTANDING}},\nYear = {{2004}},\nVolume = {{95}},\nNumber = {{3}},\nPages = {{334-353}},\nMonth = {{SEP}},\nAbstract = {{We present a novel ``dynamic learning{''} approach for an intelligent\n image database system to automatically improve object segmentation and\n labeling without user intervention, as new examples become available,\n for object-based indexing. The proposed approach is an extension of our\n earlier work on ``learning by example,{''} which addressed labeling of\n similar objects in a set of database images based on a single example.\n The proposed dynamic learning procedure utilizes multiple example object\n templates to improve the accuracy of existing object segmentations and\n labels. Multiple example templates may be images of the same object from\n different viewing angles, or images of related objects. This paper also\n introduces a new shape similarity metric called normalized area of\n symmetric differences (NASD), which has desired properties for use in\n the proposed ``dynamic learning{''} scheme, and is more robust against\n boundary noise that results from automatic image segmentation.\n Performance of the dynamic learning procedures has been demonstrated by\n experimental results. (C) 2004 Elsevier Inc. All rights reserved.}},\nDOI = {{10.1016/j.cviu.2004.04.003}},\nISSN = {{1077-3142}},\nUnique-ID = {{ISI:000223379500004}},\n}\n\n","author_short":["Xu, Y.","Saber, E","Tekalp, A."],"key":"ISI:000223379500004","id":"ISI:000223379500004","bibbaseid":"xu-saber-tekalp-dynamiclearningfrommultipleexamplesforsemanticobjectsegmentationandsearch-2004","role":"author","urls":{},"metadata":{"authorlinks":{}}},"search_terms":["dynamic","learning","multiple","examples","semantic","object","segmentation","search","xu","saber","tekalp"],"keywords":[],"authorIDs":[],"dataSources":["P7SB4qiBxZPhjXYRW","ziFHh7RJJaJNc9iie"]}