montes inaoep

Information about montes inaoep

Published on November 28, 2007

Author: Bernadette

Source: authorstream.com

Content

INAOE at ImageCLEF2007 Towards Annotation based Image Retrieval :  INAOE at ImageCLEF2007 Towards Annotation based Image Retrieval H. Jair Escalante, Carlos Hernández, Aurelio López, Heidi Marín, Manuel Montes, Eduardo Morales, Enrique Sucar, Luis Villaseñor Language Technologies Laboratory National Institute of Astrophysics, Optics and Electronics Tonantzintla, Mexico [email protected] http://ccc.inaoep.mx/~mmontesg Overview of the talk:  Overview of the talk Our first participation at ImageCLEF; the goal was to build the basic infrastructure Some textual and mixed strategies for image retrieval However we could do something more… A Web based query expansion method, and An annotation based image retrieval approach Textual and mixed strategies:  Textual and mixed strategies VSM IR System for textual retrieval (baseline) Late fusion of independent retrievers (LF) Intermedia feedback (IMFB) TBIR CBIR Fusion Relevant Images Query Example images Topic statement Some new things…:  Some new things… Web-based query expansion: Original statement + top-k snippets (NQE) Original statement + top-l more repeated words from the top-k snippets (WQE) Annotation based expansion (ABE) Use automatic image annotation methods for obtaining text from images, then… Expand documents and/or queries with automatic annotations, finally… Apply some strategy for textual image retrieval Basis of our idea:  Basis of our idea sky palm clouds sea sand sand grass palm, sky, sand, grass, sea, clouds Flamingo Beach Original name in Portuguese: “Praia do Flamengo”; Flamingo Beach is considered as one of the most beautiful beaches of Brazil; Flamingo Beach Original name in Portuguese: “Praia do Flamengo”; Flamingo Beach is considered as one of the most beautiful beaches of Brazil; Flamingo Beach Original name in Portuguese: “Praia do Flamengo”; Flamingo Beach is considered as one of the most beautiful beaches of Brazil; Region-level annotations are generally complementary to manual (image-level) annotations Automatic image annotation:  Automatic image annotation Assign labels (words) to regions within segmented images Automatic image Annotation method Grass 0.6 Sky 0.2 Tree 0.1 Ground 0.1 Grass 0.5 Tree 0.3 Ground 0.1 Jet 0.1 Rock 0.5 Church 0.2 Elephant 0.2 Entrance 0.1 Elephant Grass Sky . . . Annotation improvement Improving the automatic annotation:  Improving the automatic annotation Grass 0.6 Tree 0.2 rock 0.1 building 0.1 People 0.4 Tree 0.3 Mountain 0.2 Jet 0.1 Church 0.3 Grass 0.3 Sky 0.2 Elephant 0.2 Tree 0.5 Grass 0.3 Sky 0.1 Jet 0.1 Grass, Tree, Rock, Building, People, Mountain, Jet, Sky, Church, Elephant Set of labels:  Set of labels Some problems with the labels:  Some problems with the labels 2000 training annotated-regions (2%) 98000 regions to annotate (98%) Imbalanced training set Limited vocabulary Annotation based query expansion:  Annotation based query expansion Annotation based document expansion:  The volcano Tungurahua Baños, Ecuador March 2002 sand clouds sky mountain The surroundings of the Valle Francés Torres del Paine National Park, Chile March 2002 furniture grass people clouds Annotation based document expansion Experimental results:  Experimental results Top ranked runs for each configuration considered. Visual-English run :  Visual-English run No textual query was used, but at the end the recovery was done based on textual data. It combines intermedia feedback and our annotation based expansion technique. Textual vs. mixed strategies:  Textual vs. mixed strategies Initial conclusions:  Initial conclusions Intermedia feedback is an effective way for mixing visual and textual information Methods based on web-query expansion showed better performance Anotation based expansion is a promising way for expanding text using image’s visual content Annotations can be useful for image retrieval, though several issues should be addressed Our current work:  Our current work Work on the improvement of automatic image annotation methods Investigate different (better) ways for measuring the semantic cohesion between labels and manual annotations Use such semantic cohesion estimates for improving image retrieval from annotated collections Thanks for your attention:  Thanks for your attention Language Technologies Laboratory National Institute of Astrophysics, Optics and Electronics Tonantzintla, México Manuel Montes y Gómez [email protected] http://ccc.inaoep.mx/~mmontesg

Related presentations


Other presentations created by Bernadette

Article Usage
16. 11. 2007
0 views

Article Usage

halloween safety
05. 11. 2007
0 views

halloween safety

Oral Cancer
04. 01. 2008
0 views

Oral Cancer

service
03. 10. 2007
0 views

service

AMATYC 2004 Orlando Talk
06. 12. 2007
0 views

AMATYC 2004 Orlando Talk

16 graph design2
05. 11. 2007
0 views

16 graph design2

Astro105 Lecture13
14. 11. 2007
0 views

Astro105 Lecture13

aafsurvey 2006
16. 11. 2007
0 views

aafsurvey 2006

The Real Water World
01. 01. 2008
0 views

The Real Water World

Asteroid Mining
02. 01. 2008
0 views

Asteroid Mining

State Space Intro show
03. 01. 2008
0 views

State Space Intro show

Dali parteI
05. 01. 2008
0 views

Dali parteI

peer relationship2
28. 12. 2007
0 views

peer relationship2

ladisch purdue
04. 01. 2008
0 views

ladisch purdue

TeamSystem2008Overvi ew
01. 12. 2007
0 views

TeamSystem2008Overvi ew

2005 4160s2 06 Fasano
04. 10. 2007
0 views

2005 4160s2 06 Fasano

GM Presentation
15. 11. 2007
0 views

GM Presentation

digitisation and data capture
27. 02. 2008
0 views

digitisation and data capture

patryk
19. 12. 2007
0 views

patryk

PP O2Dock
29. 02. 2008
0 views

PP O2Dock

Editorial Policy and Process
04. 12. 2007
0 views

Editorial Policy and Process

costianes
05. 03. 2008
0 views

costianes

Export Presentation
14. 03. 2008
0 views

Export Presentation

2 Poster 5 Grant
21. 12. 2007
0 views

2 Poster 5 Grant

recitationreviewv2
13. 04. 2008
0 views

recitationreviewv2

13 1
17. 12. 2007
0 views

13 1

Podzim v Chicagu
05. 11. 2007
0 views

Podzim v Chicagu

bioterrorism new
27. 12. 2007
0 views

bioterrorism new

BlacksburgConcl2
12. 12. 2007
0 views

BlacksburgConcl2

E2005 1 347788 EteSeance5
27. 11. 2007
0 views

E2005 1 347788 EteSeance5

SONIA MENNA BARRETO 01
01. 11. 2007
0 views

SONIA MENNA BARRETO 01

Envir100LectOct29
21. 11. 2007
0 views

Envir100LectOct29

welinski 2
28. 11. 2007
0 views

welinski 2

OHP9b PCA 2006
23. 11. 2007
0 views

OHP9b PCA 2006