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 10, 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

999

D2CL: A Dual-Dimensional Contrastive Learning Method for Enhancing the Performance of Weakly Supervised Semantic

Qihang Jia1, Xiangfu Ding1, Na Tian1

  • 1School of Automation and Electronic Engineering, Qingdao University of Science and Technology, Qingdao, China.

Annals of the New York Academy of Sciences
|November 22, 2025
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

Early identification of PRISm from chest CT: a multimodal for three-class stratification of normal, PRISm, and COPD.

Frontiers in radiology·2026
Same author

Multimorbidity burden and patterns associated with DeepBrainNet-derived brain-age gap in dementia-free older adults: A community-based study.

Alzheimer's & dementia : the journal of the Alzheimer's Association·2026
Same author

Peripheral blood immune profiling reveals key signatures in newly diagnosed NK/T cell lymphoma patients.

Theranostics·2026
Same author

Habitat suitability and environmental correlates of Haemaphysalis longicornis in the central mountainous region of Shandong, China.

Acta tropica·2026
Same author

Pore Structure Engineering Enables Synergistic Interfacial-Bulk Kinetics in Hard Carbon for High-Rate Sodium-Ion Batteries.

ACS applied materials & interfaces·2026
Same author

Integrated multi-omics analysis revealed the response mechanism of <i>Osmanthus fragrans</i> leaves against the infection by <i>Botryosphaeria dothidea</i>.

Frontiers in plant science·2026

A new dual-dimensional contrastive learning (D2CL) framework enhances weakly supervised semantic segmentation by improving feature distinctiveness. This method boosts performance in autonomous driving and medical imaging by better utilizing intraclass information and separating interclass features.

Area of Science:

  • Computer Vision
  • Machine Learning
  • Artificial Intelligence

Background:

  • Weakly supervised semantic segmentation is crucial for autonomous driving and medical image analysis.
  • Current methods struggle with limited semantic cues, leading to poor intraclass representation and interclass confusion.
  • This impairs overall segmentation performance.

Purpose of the Study:

  • To introduce a novel dual-dimensional contrastive learning (D2CL) framework.
  • To enhance feature learning by exploring attributes across and within views.
  • To improve intraclass compactness and interclass separability for better segmentation.

Main Methods:

  • Proposed a dual-dimensional contrastive learning (D2CL) framework.
  • Implemented an interclass prototype contrastive learning module with a dynamic prototype memory bank.
Keywords:
contrastive learningpseudo‐labelsemantic segmentationweakly supervised learning

More Related Videos

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

3.3K

Related Experiment Videos

Last Updated: Jan 10, 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

999
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

3.3K
  • Developed an intraclass pixel contrastive learning module for pixel-wise variations within categories.

    • Main Results:

    • D2CL significantly improved the performance of baseline models on PASCAL VOC 2012 and MS COCO 2014 datasets.
    • The mean intersection over union (mIoU) for SEAM increased from 64.5% to 67.7%.
    • The mIoU for AMN improved from 69.6% to 71.8%.

    Conclusions:

    • The D2CL framework effectively enhances weakly supervised semantic segmentation.
    • The method demonstrates general applicability and consistent performance improvements across different models.
    • D2CL offers a promising approach for leveraging fine-grained feature attributes in semantic segmentation tasks.