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: Jan 12, 2026

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

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Published on: July 5, 2024

731

Single-Source Domain Defect-Aware Adaptation and Style-Modulated Generalization Network for Multispectral Image

Wei Li, Boyu Zhao, Mengmeng Zhang

    IEEE Transactions on Cybernetics
    |October 31, 2025
    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

    Research on risk assessment method of wireline logging sticking and analysis of sensitive factors.

    Scientific reports·2026
    Same author

    BSMT1, SABP3, and MAT2 assemble into a ternary complex vital for methyl salicylate biosynthesis and airborne defense.

    Molecular plant·2026
    Same author

    Combination of ALA-PDT and intracavitary triamcinolone acetonide lavage for perifolliculitis capitis abscedens et suffodiens: A case series.

    Photodiagnosis and photodynamic therapy·2026
    Same author

    Alliance: All-in-One Spectral-Spatial-Frequency Awareness Foundation Model.

    IEEE transactions on pattern analysis and machine intelligence·2025
    Same author

    Green solvent-processed n-type fluorinated salt/single-walled carbon nanotube film with high performance for multifunctional wearable thermoelectric devices.

    Journal of colloid and interface science·2025
    Same author

    Screening and prioritization of endocrine-disrupting chemicals in plastic toys for children based on non-target analysis and machine learning predictive model.

    Environment international·2025
    Same journal

    A New Human-Likeness and Comfort Index for Robot Movements Along Prescribed Paths.

    IEEE transactions on cybernetics·2026
    Same journal

    Robust Semiglobal and Global Stabilization for Nonlinear Normal Form Systems by Time-Varying Feedback.

    IEEE transactions on cybernetics·2026
    Same journal

    Adaptive Global Asymptotic Output Stabilization of Uncertain Nonlinear Systems Under Dynamic State/Input Quantization.

    IEEE transactions on cybernetics·2026
    Same journal

    Accelerated Distributed Gradient Tracking for Constrained Aggregative Optimization Over Time-Varying Digraphs.

    IEEE transactions on cybernetics·2026
    Same journal

    Small-Gain-Based Plug-and-Play Distributed Control Framework for DC Microgrids With Decentralized Reconfiguration.

    IEEE transactions on cybernetics·2026
    Same journal

    Prescribed-Time Impulsive Control of High-Order Integrator Systems.

    IEEE transactions on cybernetics·2026
    See all related articles

    This study introduces SDSnet for multispectral remote sensing image segmentation, overcoming data limitations with defect-aware prompt learning and style generalization. It achieves superior performance compared to existing methods.

    Area of Science:

    • Computer Science
    • Remote Sensing
    • Artificial Intelligence

    Background:

    • Multispectral remote sensing image (MSI) semantic segmentation suffers from limited labeled data and scene variability.
    • Existing domain adaptation (DA) and domain generalization (DG) methods have limitations, such as requiring target domain data or having restricted task adaptability.
    • The Segment Anything Model (SAM) shows promise but is not directly applicable to MSI due to its training data and prompt requirements.

    Purpose of the Study:

    • To propose a novel network, SDSnet, for effective MSI semantic segmentation.
    • To address the challenges of limited data and domain variability in MSI segmentation.
    • To improve cross-domain adaptability and inference efficiency.

    Main Methods:

    • Developed a single-source domain defect-aware adaptation and style-modulated generalization network (SDSnet).

    More Related Videos

    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

    3.3K

    Related Experiment Videos

    Last Updated: Jan 12, 2026

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

    Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

    Published on: July 5, 2024

    731
    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

    3.3K
  • Integrated defect-aware prompt learning using entropy-based defect detection to focus on challenging regions.
  • Employed style generalization learning with codebook-based style modulation for enhanced cross-domain adaptability.
  • Utilized knowledge distillation for efficient inference with the base network.

    • Main Results:

    • SDSnet demonstrated superior performance on three target domains compared to state-of-the-art DA, DG, and SAM-based methods.
    • The proposed defect-aware prompt learning effectively identifies and focuses on high-difficulty segmentation regions.
    • Style generalization learning significantly improved cross-domain adaptability.

    Conclusions:

    • SDSnet offers an effective solution for MSI semantic segmentation, particularly in scenarios with limited labeled data.
    • The network achieves efficient inference without additional computational overhead.
    • The approach shows significant advancements over existing methods, including SAM-based techniques.