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

Toward objective evaluation of image segmentation algorithms.

Ranjith Unnikrishnan1, Caroline Pantofaru, Martial Hebert

  • 1Robotics Institute, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA 15213, USA. ranjith@cs.cmu.edu

IEEE Transactions on Pattern Analysis and Machine Intelligence
|April 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

Strategies and manoeuvres for severe rigid scoliosis.

Journal of clinical orthopaedics and trauma·2025
Same author

Multifocal Skeletal Tuberculosis - A Case Report.

Journal of orthopaedic case reports·2024
Same author

Linear RGB-D SLAM for Structured Environments.

IEEE transactions on pattern analysis and machine intelligence·2021
Same author

A Large-Scale Virtual Dataset and Egocentric Localization for Disaster Responses.

IEEE transactions on pattern analysis and machine intelligence·2021
Same author

Blending of brain-machine interface and vision-guided autonomous robotics improves neuroprosthetic arm performance during grasping.

Journal of neuroengineering and rehabilitation·2016
Same author

Occlusion Reasoning for Object Detection under Arbitrary Viewpoint.

IEEE transactions on pattern analysis and machine intelligence·2015
Same journal

Relation DETR+: Exploring Explicit Position Relation Prior for Dense Prediction.

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

RBF++: Quantifying and Optimizing Reasoning Boundaries across Measurable and Unmeasurable Capabilities for Chain-of-Thought Reasoning.

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

CAFE: Cross-View Adaptive Fusion and Cluster Center Enhancement for Robust Multi-View Clustering.

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

DIVER: Reinforced Diffusion Breaks Imitation Bottlenecks in End-to-End Autonomous Driving.

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

Ethics-Aware Safe Reinforcement Learning for Rare-Event Risk Control in Interactive Urban Driving.

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

Learning Shape Anchors for Holistic Indoor Scene Understanding.

IEEE transactions on pattern analysis and machine intelligence·2026
See all related articles

This study introduces a quantitative method for evaluating unsupervised image segmentation algorithms. The Normalized Probabilistic Rand (NPR) index provides objective comparisons using ground-truth segmentations.

Area of Science:

  • Computer Vision
  • Image Processing
  • Machine Learning

Background:

  • Unsupervised image segmentation is crucial for image understanding but lacks objective evaluation metrics.
  • Current methods rely on subjective assessments of example segmented images.
  • The ill-defined nature of image segmentation hinders direct comparison with ground truth.

Purpose of the Study:

  • To introduce a quantitative evaluation framework for unsupervised image segmentation algorithms.
  • To demonstrate the utility of the Normalized Probabilistic Rand (NPR) index for this purpose.
  • To enable principled comparisons between different segmentation algorithms and across various images.

Main Methods:

  • Utilized the Normalized Probabilistic Rand (NPR) index, a measure of similarity.

Related Experiment Videos

  • Employed a hand-labeled set of ground-truth segmentations for quantitative comparison.
  • Evaluated familiar algorithms: mean-shift, graph-based, hybrid, and expectation maximization.
  • Tested on the Berkeley Segmentation Data Set comprising 300 images.

    • Main Results:

    • The NPR index facilitates objective and quantitative comparisons of segmentation algorithms.
    • Demonstrated effective evaluation across different algorithms and image datasets.
    • Results provide a basis for selecting and improving segmentation techniques.

    Conclusions:

    • The Normalized Probabilistic Rand (NPR) index offers a robust solution for the objective evaluation of unsupervised image segmentation.
    • This quantitative approach addresses the limitations of subjective assessments.
    • The proposed method supports advancements in image understanding and computer vision systems.