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

One-shot learning of object categories.

Li Fei-Fei1, Rob Fergus, Pietro Perona

  • 1University of Illinois Urbana-Champaign, 405 N. Mathews Ave., MC 251, Urbana, IL 61801, USA. feifeili@uiuc.edu

IEEE Transactions on Pattern Analysis and Machine Intelligence
|March 29, 2006
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

A framework of digital biomarkers for neurodegenerative diseases.

Nature reviews bioengineering·2026
Same author

The Language of Motion: Unifying Verbal and Non-verbal Language of 3D Human Motion.

Proceedings. IEEE Computer Society Conference on Computer Vision and Pattern Recognition·2026
Same author

Discovering Latent Graphs with GFlowNets for Diverse Conditional Image Generation.

Advances in neural information processing systems·2026
Same author

Repurposing 2D Diffusion Models with Gaussian Atlas for 3D Generation.

... IEEE International Conference on Computer Vision workshops. IEEE International Conference on Computer Vision·2026
Same author

EchoAtlas: A Conversational, Multi-View Vision-Language Foundation Model for Echocardiography Interpretation and Clinical Reasoning.

medRxiv : the preprint server for health sciences·2026
Same author

Developing ICU Clinical Behavioral Atlas Using Ambient Intelligence and Computer Vision.

NEJM AI·2026

This study introduces a Bayesian approach for learning visual object categories with minimal data. It leverages prior knowledge from diverse categories, enabling effective learning from just one or a few images, outperforming traditional methods.

Area of Science:

  • Computer Vision
  • Machine Learning
  • Artificial Intelligence

Background:

  • Learning visual models for object categories typically demands extensive datasets (hundreds or thousands of images).
  • Existing methods struggle with limited training examples, hindering the development of robust visual models.
  • There is a need for efficient learning strategies that can generalize from scarce data.

Purpose of the Study:

  • To demonstrate that significant information about object categories can be learned from very few training examples.
  • To explore a Bayesian implementation for transferring knowledge across diverse categories.
  • To compare the proposed Bayesian approach against Maximum Likelihood (ML) and Maximum A Posteriori (MAP) methods.

Main Methods:

  • Utilizing a Bayesian framework where object categories are represented by probabilistic models.

Related Experiment Videos

  • Representing prior knowledge as a probability density function on model parameters.
  • Updating prior knowledge with new observations to form posterior models for object categories.

    • Main Results:

    • A simple Bayesian algorithm was tested on a database of 101 diverse object categories.
    • The Bayesian approach successfully generated informative category models with limited training data.
    • Compared to ML and MAP methods, the Bayesian approach was more effective when training examples were scarce.

    Conclusions:

    • The proposed Bayesian method enables effective learning of visual object categories even with minimal data.
    • Transferring knowledge from previously learned categories is a key factor in achieving efficient learning.
    • This approach offers a viable solution for scenarios where large annotated datasets are unavailable.