A Linear-Algebraic Technique with
an Application in Semantic Image
Retrieval
International Conference on Image andVideo Retrieval 2006
Jonathon S. Hare and Paul H. Lewis
Intelligence,Agents, Multimedia Group
School of Electronics and Computer Science
University of Southampton
{jsh2 | phl}@ecs.soton.ac.uk
&
Peter G.B. Enser and Christine J. Sandom
School of Computing, Mathematical and Information Sciences
University of Brighton
{p.g.b.enser | c.sandom}@bton.ac.uk
Contents
Introduction
Using Linear Algebra to
Associate Images and Terms
A Simple Semantic-Space
foo
Key
SKY
TREE
MOUNTAIN
CABLE CAR
Visual terms
Keywords
Documents
Experimental Results
Real World Applications
In “The Bridging of the Semantic Gap inVisual
Information Retrieval” project we are exploring
how test-bed ontologies combined with content-
based techniques and annotation can help meet
the needs of real users in limited domains.
In particular, we are investigating how the
factorisation technique works with real image
collections.
Real World Applications
The Kennel Club Data-set
Images of dog related activities from the Kennel Club.
About 3000 annotated images (noisy keywords).
~3600 unannotated images.
Images indexed with quantised DoG/SIFT features.
3000 term vocabulary, trained on Washington
data-set.
Naively applied the factorisation technique, without
any cleaning of the keywords.
Real World Applications
The Kennel Club Data-set :: Demo

A Linear-Algebraic Technique with an Application in Semantic Image Retrieval

  • 1.
    A Linear-Algebraic Techniquewith an Application in Semantic Image Retrieval International Conference on Image andVideo Retrieval 2006 Jonathon S. Hare and Paul H. Lewis Intelligence,Agents, Multimedia Group School of Electronics and Computer Science University of Southampton {jsh2 | phl}@ecs.soton.ac.uk & Peter G.B. Enser and Christine J. Sandom School of Computing, Mathematical and Information Sciences University of Brighton {p.g.b.enser | c.sandom}@bton.ac.uk
  • 2.
  • 3.
  • 4.
    Using Linear Algebrato Associate Images and Terms
  • 5.
  • 6.
  • 7.
    Real World Applications In“The Bridging of the Semantic Gap inVisual Information Retrieval” project we are exploring how test-bed ontologies combined with content- based techniques and annotation can help meet the needs of real users in limited domains. In particular, we are investigating how the factorisation technique works with real image collections.
  • 8.
    Real World Applications TheKennel Club Data-set Images of dog related activities from the Kennel Club. About 3000 annotated images (noisy keywords). ~3600 unannotated images. Images indexed with quantised DoG/SIFT features. 3000 term vocabulary, trained on Washington data-set. Naively applied the factorisation technique, without any cleaning of the keywords.
  • 9.
    Real World Applications TheKennel Club Data-set :: Demo