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: Jul 17, 2026

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments
03:31

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments

Published on: December 15, 2023

Prototype-Anchored Generalized Manifold Regression for Unknown-Domain Object Detection.

Zihao Zhang, Aming Wu, Yang Li

    IEEE Transactions on Pattern Analysis and Machine Intelligence
    |July 15, 2026
    PubMed
    Summary

    This study introduces Manifold Regression with Visual-Text Dual Chain-of-Thought (MR-DCoT) for robust Single-Domain Generalized Object Detection (Single-DGOD). The novel approach enhances model generalization by rectifying deviant samples onto a stable semantic manifold, improving performance across diverse, unseen domains.

    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

    Molecular ground-state conformation prediction based on the Mamba state space model.

    Communications chemistry·2026
    Same author

    A dual-pronged strategy: multi-bioactive nanoplatform synergizes ROS antidote and iron stabilization to protect against liver ischemia-reperfusion injury.

    Journal of nanobiotechnology·2026
    Same author

    Effects of whole lotus root powder particle size on gel properties and microstructure of duck emulsified sausage: Underlying mechanisms of gelling behavior.

    Food chemistry·2026
    Same author

    A natural remedy: Kidney-targeting nanoplatform for photoacoustic imaging-guided three-in-one theranostics of rhabdomyolysis-induced acute kidney injury.

    Materials today. Bio·2026
    Same author

    Interaction and formation mechanism of zein-yeast carboxymethyl glucan nanoparticles for the encapsulation of curcumin.

    International journal of biological macromolecules·2026
    Same author

    Fourier-KAN: Feature Distribution Decomposition and Recombination for Unknown-Domain Object Detection.

    IEEE transactions on image processing : a publication of the IEEE Signal Processing Society·2026

    Area of Science:

    • Computer Vision
    • Machine Learning
    • Artificial Intelligence

    Background:

    • Single-Domain Generalized Object Detection (Single-DGOD) aims to transfer models to multiple unknown domains.
    • Current methods using simulation-driven paradigms struggle with real-world dynamic variations, leading to overfitting and poor handling of structural degradations.
    • The manifold hypothesis suggests semantic features lie on a compact manifold, implying generalization requires rectifying off-manifold samples.

    Purpose of the Study:

    • To propose a novel framework, Manifold Regression with Visual-Text Dual Chain-of-Thought (MR-DCoT), for robust Single-Domain Generalized Object Detection.
    • To address the limitations of existing methods in capturing real-world variations and handling structural degradations.
    • To improve model generalization and robustness to unseen domain shifts.

    Related Experiment Videos

    Last Updated: Jul 17, 2026

    End-To-End Deep Neural Network for Salient Object Detection in Complex Environments
    03:31

    End-To-End Deep Neural Network for Salient Object Detection in Complex Environments

    Published on: December 15, 2023

    Main Methods:

    • Reformulated robust generalization as a manifold regression problem.
    • Developed a Visual-Text Dual Chain-of-Thought module coupling VLM-guided semantics and diffusion-based perturbations to generate hard examples.
    • Introduced Class-Specific Prototype Anchoring to learn a rectification operator guiding features back to the source semantic manifold.

    Main Results:

    • The MR-DCoT framework effectively bridges the distribution gap between source and target domains.
    • Demonstrated significant improvements in generalization and robustness to unseen shifts.
    • Achieved superior performance on diverse benchmarks, including varying weather conditions, real-to-art generalization, and zero-shot semantic segmentation.

    Conclusions:

    • The proposed MR-DCoT framework offers a versatile and effective solution for Single-Domain Generalized Object Detection.
    • Learning to rectify deviant samples onto a stable semantic manifold is key to achieving robust generalization.
    • The method shows strong potential for real-world applications requiring reliable object detection across diverse and challenging environments.