Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Supervised learning of semantic classes for image annotation and retrieval.

Gustavo Carneiro1, Antoni B Chan, Pedro J Moreno

  • 1Integrated Data Systems Department, Siemens Corporate Research, Princeton, NJ 08540, USA. gustavo.carneiro@siemens.com

IEEE Transactions on Pattern Analysis and Machine Intelligence
|January 17, 2007
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

AutoCumulus: an automated mammographic density measure created using artificial intelligence.

BMC cancer·2026
Same author

Deep learning Algorithm for Wound assessment after total kNee (DAWN) arthroplasty : a prospective study protocol.

Bone & joint open·2026
Same author

Rapid Artificial Intelligence Autoplanning Rivals Manual Expert Planning for Cervical Brachytherapy.

Practical radiation oncology·2026
Same author

Macula Spatial Patterns and Their Association With Central Visual Field Progression in Glaucoma Using Artificial Intelligence.

Journal of glaucoma·2026
Same author

Bridging Generative and Discriminative Noisy-Label Learning via Direction-Agnostic EM Formulation.

IEEE transactions on pattern analysis and machine intelligence·2026
Same author

Metabolomic profiling in phenylketonuria: a systematic review of human studies.

Metabolomics : Official journal of the Metabolomic Society·2026

This study introduces a simple, efficient probabilistic method for semantic image annotation and retrieval. The approach achieves high accuracy with less computation than existing techniques.

Area of Science:

  • Computer Vision
  • Machine Learning
  • Probabilistic Modeling

Background:

  • Semantic image annotation and retrieval are crucial for organizing and accessing large image datasets.
  • Current methods often involve complex joint modeling of visual features and semantic labels, requiring significant computational resources and prior image segmentation.

Purpose of the Study:

  • To propose a novel probabilistic formulation for semantic image annotation and retrieval.
  • To develop algorithms that are conceptually simple, computationally efficient, and do not require prior semantic segmentation.
  • To demonstrate the superiority of a supervised formulation over complex joint modeling approaches.

Main Methods:

  • Images are represented as bags of localized feature vectors.
  • A mixture density is estimated for each image.

Related Experiment Videos

  • Densities for images with common semantic labels are pooled into semantic class estimates using a hierarchical Expectation-Maximization algorithm, justified by multiple instance learning.

    • Main Results:

    • The proposed supervised formulation achieves a minimum probability of error in annotation and retrieval.
    • It demonstrates higher accuracy compared to various previously published methods.
    • The method operates at a fraction of the computational cost of existing techniques.

    Conclusions:

    • The developed probabilistic formulation offers a conceptually simple and computationally efficient solution for semantic image annotation and retrieval.
    • This supervised approach outperforms complex joint modeling methods in terms of accuracy and efficiency.
    • The method shows robustness to parameter tuning, making it practical for real-world applications.