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 Videos

SynerNet: Broad-to-precise CAM synergy for weakly supervised semantic segmentation.

Zhonggai Wang1, Guangyu Gao1, Zhuoshu Li1

  • 1School of Computer Science and Technology, Beijing Institute of Technology, Beijing, China.

Neural Networks : the Official Journal of the International Neural Network Society
|May 6, 2026
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

FGF21 rejuvenates aged human adipose-derived mesenchymal stem cells via enhancement of TFE3-mediated autophagy flux.

Autophagy·2026
Same author

Growth arrest-specific 6 rejuvenates senescent HUCMSCs through upregulating Nrf2 for diabetic wound therapy.

Free radical biology & medicine·2026
Same author

Associations of changes in frailty with multiple chronic diseases in middle aged and elderly adults.

BMC public health·2025
Same author

AutoSGRL: Automated framework construction for self-supervised graph representation learning.

Neural networks : the official journal of the International Neural Network Society·2025
Same author

DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning.

Nature·2025
Same author

Genetic Evidence on the Role of Blood Phytosterols in Frailty: A Mendelian Randomization Study.

Food science & nutrition·2025
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

SynerNet, a novel one-stage dual-branch framework, enhances weakly supervised semantic segmentation by generating both broad and precise pseudo-labels. This approach overcomes limitations of traditional methods, achieving state-of-the-art results.

Area of Science:

  • Computer Vision
  • Deep Learning
  • Image Segmentation

Background:

  • Weakly Supervised Semantic Segmentation (WSSS) is challenging due to noisy image-level labels.
  • One-stage WSSS methods suffer from encoder-label coupling, reinforcing initial inaccuracies.
  • Existing approaches struggle with incomplete or imprecise class activation maps (CAMs).

Purpose of the Study:

  • To introduce SynerNet, a one-stage dual-branch framework for improved WSSS.
  • To address the limitations of error propagation in coupled encoder-decoder architectures.
  • To generate high-fidelity pseudo-labels that are both comprehensive and accurate.

Main Methods:

  • Employed a dual-branch architecture with complementary objectives: B-CAM for broad coverage and P-CAM for precise localization.
Keywords:
CAMCo-trainingDual-branchPseudo-labelWeakly supervised semantic segmentation

Related Experiment Videos

  • Utilized global attention in B-CAM to expand foreground regions and emphasized background in P-CAM for localization.
  • Implemented cross-supervision between branches to decouple optimization and mitigate error reinforcement.
  • Introduced a confidence matrix from multi-scale ViT features for confidence-guided fusion of predictions.

    • Main Results:

    • SynerNet achieved state-of-the-art performance on PASCAL VOC 2012 and COCO 2014 datasets.
    • The dual-branch approach successfully generated pseudo-labels that were both complete and precise.
    • Demonstrated effective mitigation of error reinforcement through decoupled optimization and confidence-guided fusion.

    Conclusions:

    • SynerNet offers an effective one-stage co-training strategy for high-quality WSSS.
    • The proposed complementary objectives and confidence-guided fusion significantly improve segmentation accuracy.
    • The framework successfully generates high-fidelity pseudo-labels in an end-to-end manner.