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: Mar 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

852

WEViT: weight-entangled vision transformers with class-specific attention for weakly supervised semantic

Narges Saeedizadeh1, Seyed Mohammad Jafar Jalali2, Burhan Khan1

  • 1Institute for Intelligent Systems Research and Innovation, Deakin University, Geelong, Australia.

Neural Networks : the Official Journal of the International Neural Network Society
|March 11, 2026
PubMed
Summary

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

Redox-mediator enhanced electrochemiluminescence under non-aqueous conditions.

Chemical science·2026
Same author

Automatic assessment of lung involvement in systemic sclerosis using deep learning.

Journal of research in medical sciences : the official journal of Isfahan University of Medical Sciences·2026
Same author

BAOS-CNN: A novel deep neuroevolution algorithm for multispecies seagrass detection.

PloS one·2024
Same author

Biased Deep Learning Methods in Detection of COVID-19 Using CT Images: A Challenge Mounted by Subject-Wise-Split ISFCT Dataset.

Journal of imaging·2023
Same author

Magnetic resonance imaging of pilonidal sinus disease: interobserver agreement and practical MRI reporting tips.

European radiology·2023
Same author

A virtual reality study investigating the train illusion.

Royal Society open science·2023
Same journal

Exploiting audio-visual modalities in videos: Object detection via multi-stage bilateral coupling network.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Reliability-aware modality completion with cross-modal distillation for federated learning with missing modalities.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

IGFD-Net: Illumination-guided frequency decoupling for polarization image fusion.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Multiple-Strategies dung beetle optimizer and its applications in engineering optimization and bankruptcy prediction.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Aggregating global-scale pixel-wise forgery cues within a graph.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Finite-Time intermittent control for secure synchronization of Neutral-Type stochastic delayed neural networks under aperiodic DoS attacks.

Neural networks : the official journal of the International Neural Network Society·2026
See all related articles
This summary is machine-generated.

This study introduces WEViT, a novel framework integrating Neural Architecture Search (NAS) with transformers for Weakly Supervised Semantic Segmentation (WSSS). WEViT optimizes network architectures for accurate object localization, achieving state-of-the-art results.

Area of Science:

  • Computer Vision
  • Deep Learning

Background:

  • Weakly Supervised Semantic Segmentation (WSSS) traditionally uses Class Activation Maps (CAMs) but faces challenges in balancing localization accuracy and scalability.
  • Existing methods often rely on fixed network architectures and manual strategies, limiting adaptability.
  • Neural Architecture Search (NAS) has not been applied to WSSS due to the need for efficient weight sharing.

Purpose of the Study:

  • To propose WEViT, a novel framework combining NAS with transformers for WSSS.
  • To optimize network architectures for generating accurate, class-specific object localization maps.
  • To address the limitations of traditional WSSS methods by improving scalability and efficiency.

Main Methods:

  • WEViT integrates NAS with transformers, utilizing a weight entanglement strategy for efficient supernet training and weight inheritance.
Keywords:
Class activation mappingClass tokenEvolutionaryNeural architecture searchObject localizationSearch spaceTransformerWeakly supervised semantic segmentation

More Related Videos

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
04:23

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

Published on: April 21, 2023

2.4K

Related Experiment Videos

Last Updated: Mar 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

852
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
04:23

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

Published on: April 21, 2023

2.4K
  • An evolutionary algorithm selects the optimal architecture, from which transformer head attention weights are extracted.
  • A Refinement Patch Affinity strategy and a regularization loss function are employed to enhance localization accuracy and class discrimination.

    • Main Results:

    • WEViT achieves state-of-the-art performance on benchmark datasets like PASCAL VOC 2012 and MS COCO.
    • The framework demonstrates the effectiveness of applying NAS to WSSS for the first time.
    • The weight entanglement strategy significantly reduces computational costs by avoiding subnet retraining.

    Conclusions:

    • WEViT offers a scalable, efficient, and accurate solution for Weakly Supervised Semantic Segmentation.
    • The integration of NAS and transformers represents a significant advancement in WSSS.
    • This work opens new avenues for optimizing deep learning architectures in segmentation tasks.