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: Jun 13, 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

Scene-aware contrastive learning with multi-scale domain-invariant feature alignment for robust visual recognition

Wenjun Li1

  • 1College of Artificial Intelligence, Putian University, Putian, 351100, Fujian, China. KKms112233@163.com.

Scientific Reports
|June 11, 2026
PubMed
Summary

Related Concept Videos

Force Classification01:22

Force Classification

Forces play a crucial role in the study of physics and engineering. They are essential in describing the motion, behavior, and equilibrium of objects in the physical world. Forces can be classified based on their origin, type, and direction of action.
Contact and non-contact forces are two of the most widely used categories of forces. As the name suggests, contact forces require physical contact between two objects to act upon each other. Examples of contact forces include frictional,...

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

TOF-SIMS surface analysis of chemical components of size-fractioned urban aerosols in a typical heavy air pollution event in Beijing.

Journal of environmental sciences (China)·2018
Same author

Interrogation of nonconserved human adipose lincRNAs identifies a regulatory role of <i>linc-ADAL</i> in adipocyte metabolism.

Science translational medicine·2018
Same author

Dietary Long-Chain n-3 Fatty Acid Intake and Arthritis Risk in the Women's Health Initiative.

Journal of the Academy of Nutrition and Dietetics·2018
Same author

[Micronutrients intake of pregnant women from 8 cities of China during 2011 to 2012].

Wei sheng yan jiu = Journal of hygiene research·2018
Same author

Phosphorylation of Parkin at serine 131 by p38 MAPK promotes mitochondrial dysfunction and neuronal death in mutant A53T α-synuclein model of Parkinson's disease.

Cell death & disease·2018
Same author

Association of Physical Activity with Late-life Mobility Limitation among Women with Total Joint Replacement for Knee or Hip Osteoarthritis.

The Journal of rheumatology·2018

This study introduces a robust deep learning framework for visual recognition in challenging conditions. The method enhances model performance across different domains and adverse scenarios without needing target data.

Area of Science:

  • Computer Vision
  • Machine Learning
  • Deep Learning

Background:

  • Deep learning models struggle with real-world visual variations like changing illumination, occlusions, and weather.
  • Existing methods often fail when deployed in complex, uncontrolled environments due to domain shift.

Purpose of the Study:

  • To develop an end-to-end framework enhancing representational robustness and domain-invariant transfer for deep learning models.
  • To improve model generalization across diverse visual domains and challenging conditions.

Main Methods:

  • A scene-aware contrastive learning strategy with physically grounded transformations and adaptive difficulty weighting.
  • Multi-scale domain-invariant feature alignment with cross-scale consistency regularization.
  • The framework operates without target-domain data and with minimal annotation.
Keywords:
Complex visual scenesContrastive learningDomain generalizationMulti-scale feature alignmentScene-aware augmentationSelf-supervised learning

Related Experiment Videos

Last Updated: Jun 13, 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 Results:

  • Outperformed supervised, self-supervised, domain generalization, and domain adaptation baselines on PACS and BDD-City benchmarks.
  • Achieved average cross-domain accuracy gains of 3.5 and over 4% points, respectively.
  • Significant performance improvements under severe degradation conditions where baselines falter.

Conclusions:

  • The proposed framework effectively enhances deep learning model robustness and domain generalization.
  • Both proposed components contribute independently, with synergistic interaction for optimal performance.
  • The method offers a promising solution for reliable visual recognition in uncontrolled environments.