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 Video

Updated: Jun 22, 2025

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

2.7K

Source-free domain adaptation for semantic image segmentation using internal representations.

Serban Stan1, Mohammad Rostami1

  • 1Department of Computer Science, University of Southern California, Los Angeles, CA, United States.

Frontiers in Big Data
|July 3, 2024
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

An efficient triple-domain YOLOv8 for real-time aluminum profile defect detection.

Scientific reportsยท2026
Same author

Multi-level information fusion for explainable diagnosis of melanoma using dermoscopic images.

Computerized medical imaging and graphics : the official journal of the Computerized Medical Imaging Societyยท2026
Same author

Identification of defects in pure and Al/Ga-doped ZnO to improve X-ray detector performance: experimental and simulation methods.

Physical chemistry chemical physics : PCCPยท2025
Same author

In situ construction of green ZnFe<sub>2</sub>O<sub>4</sub>/sub-5nm N, Cu dual-doped SnO<sub>2</sub> S-scheme heterostructure with the boosted spatial charge separation towards decontamination of tetracycline: Mechanistic perspectives and aquatic hazard assessment.

Journal of environmental managementยท2025
Same author

In vivo validation of a smart sensor-enabled dressing for remote wound monitoring.

Biosensors & bioelectronicsยท2025
Same author

Volatile organic compounds (VOCs) detection for the identification of bacterial infections in clinical wound samples.

Talantaยท2025

This study introduces an online unsupervised domain adaptation (UDA) algorithm for semantic segmentation. It enhances model generalization on new data without needing source data access during adaptation.

Area of Science:

  • Computer Vision
  • Machine Learning
  • Artificial Intelligence

Background:

  • Semantic segmentation models struggle with performance degradation due to changing data distributions over time.
  • Re-training models is often necessary to maintain accuracy with new data.
  • Unsupervised Domain Adaptation (UDA) addresses this by transferring knowledge from labeled source domains to unlabeled target domains.

Purpose of the Study:

  • To develop an online UDA algorithm for semantic segmentation that improves model generalization.
  • To enable adaptation to unannotated target domains with restricted source data access.
  • To enhance the robustness of semantic segmentation models against data distribution shifts.

Main Methods:

  • Minimizing distributional distance between source and target latent features in a shared embedding space.
Keywords:
Gaussian mixture model (GMM)domain adaptationimage segmentationoptimal transport and Wasserstein distancessliced Wasserstein distance

More Related Videos

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
04:48

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Published on: July 5, 2024

385
A Label-Free Segmentation Approach for Intravital Imaging of Mammary Tumor Microenvironment
10:39

A Label-Free Segmentation Approach for Intravital Imaging of Mammary Tumor Microenvironment

Published on: May 24, 2022

2.3K

Related Experiment Videos

Last Updated: Jun 22, 2025

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

2.7K
Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
04:48

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Published on: July 5, 2024

385
A Label-Free Segmentation Approach for Intravital Imaging of Mammary Tumor Microenvironment
10:39

A Label-Free Segmentation Approach for Intravital Imaging of Mammary Tumor Microenvironment

Published on: May 24, 2022

2.3K
  • Promoting a domain-agnostic latent feature space for classifier generalization.
  • Approximating the source latent feature distribution using a Gaussian Mixture Model (GMM) to avoid direct source data access during adaptation.

    • Main Results:

    • The proposed online UDA algorithm enhances model generalization on target domains with different data distributions.
    • The method successfully adapts semantic segmentation models without requiring access to source data during the adaptation phase.
    • Utilizing a GMM as a surrogate for source distribution proved effective in facilitating adaptation.

    Conclusions:

    • The developed online UDA approach offers a viable solution for maintaining semantic segmentation performance in dynamic environments.
    • Restricted source data access during adaptation is effectively managed by approximating source distributions.
    • This work contributes to more robust and adaptable semantic segmentation systems in real-world applications.