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

A thousand words in a scene.

Pedro Quelhas1, Florent Monay, Jean-Marc Odobez

  • 1IDIAP Research Institute, Martigny, Switzerland. pedro.quelhas@idiap.ch

IEEE Transactions on Pattern Analysis and Machine Intelligence
|July 14, 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

UveAI: clinic-ready scoring of retinal inflammation in uveitis on widefield fluorescein angiography using AI.

Scientific reports·2026
Same author

From pixels to perception: A benchmark for human-like symmetry detection.

Vision research·2026
Same author

From Forest to Zoo: Great Ape Behavior Recognition with ChimpBehave.

International journal of computer vision·2025
Same author

Decoding community proximity discourse: A mixed-methods comparative analysis of online local and national newspapers in Romandy, Switzerland.

PloS one·2025
Same author

Surgeons versus computer vision: a comparative analysis on surgical phase recognition capabilities.

International journal of computer assisted radiology and surgery·2025
Same author

Continual pre-training mitigates forgetting in language and vision.

Neural networks : the official journal of the International Neural Network Society·2024

This study shows that a text-like "bag-of-visterms" representation effectively models visual scenes for classification. Unsupervised models like Probabilistic Latent Semantic Analysis (PLSA) offer robust scene representation and pattern discovery.

Area of Science:

  • Computer Vision
  • Machine Learning
  • Artificial Intelligence

Background:

  • Traditional scene classification methods often struggle with complex visual data.
  • The efficacy of text-modeling techniques for visual scene analysis remains an open question.
  • Unsupervised learning offers potential for discovering inherent patterns in visual data.

Purpose of the Study:

  • To evaluate a text-like bag-of-visterms representation for visual scene classification.
  • To explore analogies between text document representation and discrete scene representations.
  • To investigate the use of unsupervised latent space models for feature extraction and pattern discovery in visual scenes.

Main Methods:

  • Utilized a bag-of-visterms representation, a histogram of quantized local visual features.

Related Experiment Videos

  • Applied Probabilistic Latent Semantic Analysis (PLSA) as an unsupervised latent space model.
  • Conducted experiments on multiple datasets for binary and multi-class scene classification and image ranking.

    • Main Results:

    • The bag-of-visterms representation consistently outperformed classical scene classification approaches.
    • PLSA provided a compact, discriminative, and robust scene representation, especially with limited labeled data.
    • PLSA successfully extracted meaningful scene patterns for image collection browsing.

    Conclusions:

    • Text-modeling approaches, like bag-of-visterms, are suitable for visual scene classification.
    • Unsupervised latent space models (PLSA) are effective for scene representation, classification, and pattern discovery.
    • The proposed methods offer competitive or superior performance compared to existing complex approaches.