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 8, 2026

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
03:31

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

990

Toward Effective Model Merging in Semantic Segmentation.

Haotian Chen, Yanyu Xu, Yonghui Xu

    IEEE Transactions on Neural Networks and Learning Systems
    |December 22, 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

    Genetic modification of the marine-derived yeast Yarrowia lipolytica with high-protein content using a GPI-anchor-fusion expression system.

    Biotechnology progress·2009
    Same author

    Regulation of the Edwardsiella tarda hemolysin gene and luxS by EthR.

    Journal of microbiology and biotechnology·2009
    Same author

    EBV LMP2A-specific T cell immune responses elicited by dendritic cells loaded with LMP2A protein.

    Cellular & molecular immunology·2009
    Same author

    [Combination of volar buttress plate with external fixator for the distal radial fractures of type C3 caused by high-energy injuries].

    Zhongguo gu shang = China journal of orthopaedics and traumatology·2009
    Same author

    Environmental regulation of floral anthocyanin synthesis in Ipomoea purpurea.

    Molecular ecology·2009
    Same author

    PI3K integrates the effects of insulin and leptin on large-conductance Ca2+-activated K+ channels in neuropeptide Y neurons of the hypothalamic arcuate nucleus.

    American journal of physiology. Endocrinology and metabolism·2009
    Same journal

    Hidden Data Recovery and Forecasting via Next-Generation Reservoir Computing With Multiscale Delay Selection.

    IEEE transactions on neural networks and learning systems·2026
    Same journal

    CAFF-CIL: Causality-Aware Freedom Forgetting Approach for Class-Incremental Learning.

    IEEE transactions on neural networks and learning systems·2026
    Same journal

    Harmonic Autoencoding Framework for Multiple Tasks in Magnetic Particle Imaging Reconstruction.

    IEEE transactions on neural networks and learning systems·2026
    Same journal

    A Survey on Human-Centric Voice-Face Multimodal Learning.

    IEEE transactions on neural networks and learning systems·2026
    Same journal

    Vision-Assisted Foundation Model for Solving Multitask Vehicle Routing Problems.

    IEEE transactions on neural networks and learning systems·2026
    Same journal

    FP3O: Enabling Proximal Policy Optimization in Multiagent Cooperation With Parameter-Sharing Versatility.

    IEEE transactions on neural networks and learning systems·2026
    See all related articles

    Model merging for semantic segmentation is improved by M²Seg, which uses adaptive merging and dynamic calibration to overcome distribution shifts and enhance performance on diverse tasks.

    Area of Science:

    • Computer Vision
    • Machine Learning
    • Artificial Intelligence

    Background:

    • Model merging combines individual models for improved performance, but faces challenges in semantic segmentation.
    • Existing methods use static merging, limiting adaptation to task-specific knowledge.
    • Semantic segmentation suffers from negative transfer due to domain distribution shifts.

    Purpose of the Study:

    • To propose an effective model merging approach for semantic segmentation, named M²Seg.
    • To address limitations of static merging and distribution shifts in current model merging techniques.
    • To enhance the performance and generalization capabilities of merged models for semantic segmentation tasks.

    Main Methods:

    • Introduced a novel SVD-structured Mixture of Experts (MoE) module for adaptive merging based on input data.

    More Related Videos

    Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application
    05:56

    Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application

    Published on: April 14, 2023

    3.1K

    Related Experiment Videos

    Last Updated: Jan 8, 2026

    Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
    03:31

    Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

    Published on: December 15, 2023

    990
    Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application
    05:56

    Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application

    Published on: April 14, 2023

    3.1K
  • Developed a test-time dynamic calibration function to minimize discrepancies between training and test statistics.
  • Implemented a pixel-efficient entropy minimization mechanism to filter unstable pixels and stabilize merging.

    • Main Results:

    • M²Seg demonstrates superior effectiveness in semantic segmentation tasks.
    • The method shows enhanced generalization capabilities on both seen and unseen datasets.
    • Experimental results validate the effectiveness of adaptive merging, dynamic calibration, and entropy minimization.

    Conclusions:

    • M²Seg effectively overcomes challenges in model merging for semantic segmentation.
    • The proposed approach significantly improves performance and generalization across diverse domains.
    • M²Seg offers a robust solution for combining models in semantic segmentation applications.