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

Generic object recognition with boosting.

Andreas Opelt1, Axel Pinz, Michael Fussenegger

  • 1Institute of Electrical Measurement and Measurement Signal Processing, Graz University of Technology, Austria. opelt@tugraz.at

IEEE Transactions on Pattern Analysis and Machine Intelligence
|March 11, 2006
PubMed
Summary

This study presents a weakly supervised categorization framework using local region extraction and boosting for feature selection. The approach achieves high recognition rates, outperforming others on complex image databases.

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

Using wedge-loaded compact tension specimens for investigating hydrogen-induced cracking.

Scientific reports·2026
Same author

Frequency of use and sonority sequencing in first- and second-language consonant cluster perception: facilitation is language-specific.

Frontiers in psychology·2025
Same author

Bending behavior of hybrid laminates made of aluminum and wood for sustainable lightweight structures.

Scientific reports·2025
Same author

The vowel space of multiethnolectal (Stuttgart) German.

Phonetica·2024
Same author

Speech Errors in the Production of Initial Consonant Clusters: The Roles of Frequency and Sonority.

Journal of speech, language, and hearing research : JSLHR·2022
Same author

A Study of Word Complexity Under Conditions of Non-experimental, Natural Overt Speech Production Using ECoG.

Frontiers in human neuroscience·2022

Area of Science:

  • Computer Vision
  • Machine Learning
  • Pattern Recognition

Background:

  • Weakly supervised learning offers a flexible approach to categorization when labeled data is scarce.
  • Traditional methods often rely on specific feature extractors and descriptors, limiting their adaptability.
  • Developing robust categorization systems for complex, real-world images remains a challenge.

Purpose of the Study:

  • To explore the capabilities and constraints of weakly supervised categorization.
  • To introduce a comprehensive framework for visual categorization using adaptable feature extraction and selection.
  • To evaluate the system's performance on challenging image datasets with variations in scale, pose, clutter, and occlusion.

Main Methods:

  • Extraction of local regions based on discontinuity or homogeneity.

Related Experiment Videos

  • Application of diverse local descriptors to generate feature vectors.
  • Utilizing boosting algorithms to learn and combine weak hypotheses (feature subsets) into a final category classifier.
  • Development of complex image databases for rigorous system evaluation.

    • Main Results:

    • The proposed framework achieves superior recognition rates by combining multiple feature extractors and descriptors.
    • Classification accuracy reached up to 81 percent ROC-equal error rate on highly complex image databases.
    • The system demonstrated superior performance compared to comparable approaches on standard benchmark datasets.

    Conclusions:

    • Weakly supervised categorization, when implemented with a flexible framework, is a powerful tool for image analysis.
    • The developed system effectively handles variations in object scale, pose, background clutter, and occlusion.
    • This approach offers a robust and adaptable solution for visual categorization tasks, outperforming existing methods.