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: Oct 7, 2025

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

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Published on: July 5, 2024

546

Attentive WaveBlock: Complementarity-Enhanced Mutual Networks for Unsupervised Domain Adaptation in Person

Wenhao Wang, Fang Zhao, Shengcai Liao

    IEEE Transactions on Image Processing : a Publication of the IEEE Signal Processing Society
    |January 11, 2022
    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

    Synergistic inhibitory effect of Smo inhibitor jervine and its combination with decitabine can target Hedgehog signaling pathway to inhibit myelodysplastic syndrome cell line.

    Hematology (Amsterdam, Netherlands)·2021
    Same author

    Cardiac, Macro-, and Micro-Circulatory Abnormalities in Association With Individual Metabolic Syndrome Component: The Northern Shanghai Study.

    Frontiers in cardiovascular medicine·2021
    Same author

    PIWI-interacting RNAs: Mitochondria-based biogenesis and functions in cancer.

    Genes & diseases·2021
    Same author

    Interferon-armed RBD dimer enhances the immunogenicity of RBD for sterilizing immunity against SARS-CoV-2.

    Cell research·2021
    Same author

    Molecular evolution of bumble bee vitellogenin and vitellogenin-like genes.

    Ecology and evolution·2021
    Same author

    Hypoxia-related lncRNAs to build prognostic classifier and reveal the immune characteristics of EGFR wild type and low expression of PD-L1 squamous and adenocarcinoma NSCLC.

    Cancer medicine·2021

    This study introduces the Attentive WaveBlock (AWB) to improve unsupervised domain adaptation for person re-identification by reducing noisy pseudo-labels. The novel module enhances dual network complementarity, achieving state-of-the-art results.

    Area of Science:

    • Computer Vision
    • Machine Learning
    • Artificial Intelligence

    Background:

    • Unsupervised domain adaptation (UDA) for person re-identification faces challenges due to significant domain gaps.
    • Existing self-training methods often suffer from noisy pseudo-labels, hindering model optimization.
    • Mutual learning with dual networks aims to generate reliable soft labels but can lose complementarity.

    Purpose of the Study:

    • To propose a novel module, the Attentive WaveBlock (AWB), to enhance dual networks in mutual learning for UDA person re-identification.
    • To improve the reliability of pseudo-labels by reducing noise and enhancing feature complementarity.
    • To demonstrate the effectiveness and generalizability of the proposed method.

    Main Methods:

    • Introduced a parameter-free WaveBlock module to create feature differences between dual networks by manipulating feature map blocks.

    Related Experiment Videos

    Last Updated: Oct 7, 2025

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

    Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

    Published on: July 5, 2024

    546
  • Integrated an attention mechanism to amplify feature differences and discover complementary features.
  • Explored pre-attention and post-attention combination strategies within the dual network framework.

    • Main Results:

    • The Attentive WaveBlock (AWB) significantly improved performance on multiple UDA person re-identification tasks, achieving state-of-the-art results.
    • The method demonstrated effectiveness in reducing noise in pseudo-labels and enhancing feature complementarity.
    • The proposed approach proved generalizable, showing strong performance on vehicle re-identification and image classification tasks.

    Conclusions:

    • The Attentive WaveBlock (AWB) is an effective lightweight module for improving mutual learning in UDA person re-identification.
    • AWB enhances feature complementarity and suppresses noise, leading to superior performance.
    • The module's versatility is confirmed across various computer vision tasks.