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

Optimal linear representations of images for object recognition.

Xiuwen Liu1, Anuj Srivastava, Kyle Gallivan

  • 1Department of Computer Science, Florida State University, Tallahassee, FL 32306, USA. liux@cs.fsu.edu

IEEE Transactions on Pattern Analysis and Machine Intelligence
|October 6, 2004
PubMed
Summary

This study introduces a novel stochastic gradient algorithm to find optimal linear image representations for object recognition. The method enhances recognition performance by optimizing representations on a Grassmann manifold.

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

STITCH links cellular morphology and gene expression in spatial transcriptomics.

bioRxiv : the preprint server for biology·2026
Same author

Comparative efficacy of levosimendan and dobutamine in sepsis-related cardiac impairment: a meta-analysis.

European journal of clinical pharmacology·2026
Same author

Peak Persistence Diagrams for Shape-Based Signal Estimation.

IEEE transactions on signal processing : a publication of the IEEE Signal Processing Society·2026
Same author

A shape-based functional index for objective assessment of pediatric motor function.

PloS one·2025
Same author

Endoscopic purse-string suture technique for postoperative drain-related leakage management in gastrointestinal surgery.

BMC surgery·2025
Same author

A Wasserstein-Type Distance for Gaussian Mixtures on Vector Bundles with Applications to Shape Analysis.

SIAM journal on imaging sciences·2025

Area of Science:

  • Computer Vision
  • Machine Learning
  • Image Analysis

Background:

  • Linear representations are common in image analysis but often suboptimal for specific tasks.
  • Optimizing these representations is crucial for improving performance in applications like object recognition.

Purpose of the Study:

  • To propose a novel stochastic gradient algorithm for discovering optimal linear image representations.
  • To enhance appearance-based object recognition by maximizing a defined performance function.

Main Methods:

  • Development of a stochastic gradient algorithm utilizing intrinsic flows.
  • Optimization performed on a Grassmann manifold to find optimal linear representations.
  • Nearest neighbor classifier used to define the recognition performance function.

Related Experiment Videos

Main Results:

  • The proposed algorithm effectively finds optimal linear representations for images.
  • Experimental results demonstrate significant improvements in recognition performance.
  • The algorithm's efficacy is shown on a Grassmann manifold.

Conclusions:

  • The novel stochastic gradient algorithm offers an effective approach for optimizing linear image representations.
  • This method has strong potential for advancing appearance-based object recognition systems.
  • The use of intrinsic flows on Grassmann manifolds provides a robust optimization framework.