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: May 5, 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

1.3K

Salient Object Detection with Semantic-Aware Edge Refinement and Edge-Guided Cross-Attention Feature Aggregation.

Yitong Lu1, Ziguan Cui2

  • 1Portland Institute, Nanjing University of Posts and Telecommunications, Nanjing 210023, China.

Sensors (Basel, Switzerland)
|May 4, 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

Lightweight Spatial-Frequency Collaborative Interaction Network for RGB-D Salient Object Detection.

Sensors (Basel, Switzerland)·2026
Same author

Three-Dimensional ResNeXt Network Using Feature Fusion and Label Smoothing for Hyperspectral Image Classification.

Sensors (Basel, Switzerland)·2020
See all related articles

SECA-Net unifies CNN and Transformer features for salient object detection, effectively using edge cues to improve boundary accuracy and outperform existing methods.

Area of Science:

  • Computer Vision
  • Deep Learning
  • Image Processing

Background:

  • Hybrid multi-backbone architectures and edge cue utilization are key trends in salient object detection (SOD).
  • CNNs excel at local structures, while Transformers capture global dependencies, but feature discrepancies and boundary degradation pose challenges.
  • Effectively using edge cues without introducing false boundaries or leakages remains an open problem.

Purpose of the Study:

  • To present SECA-Net, a unified framework synergizing CNN and Transformer representations for salient object detection.
  • To address feature discrepancies between CNNs and Transformers and mitigate boundary-detail degradation during multi-scale fusion.
  • To leverage edge cues effectively for structural guidance, reducing texture-induced false contours and boundary leakages.

Main Methods:

Keywords:
cross-attentiondual-encoder saliency networkedge refinementsalient object detectionsemantic-aware

Related Experiment Videos

Last Updated: May 5, 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

1.3K
  • A dual-encoder structure extracts features, employing a level-wise feature interaction (LFI) module for discrepancy-aware fusion.
  • A semantic-aware edge refinement (SAER) module generates clean edge priors using high-level semantic guidance.
  • An edge-guided cross-attention feature aggregation (ECFA) module progressively injects refined edge priors into multi-scale decoding.

Main Results:

  • SECA-Net demonstrates superior performance against 19 state-of-the-art methods across five benchmark SOD datasets.
  • The proposed method achieved top rankings in Fβ and BDE metrics on all tested datasets.
  • A notable improvement of 1.54% in Fβ was observed on the DUTS-TE dataset.

Conclusions:

  • SECA-Net effectively bridges CNN-Transformer discrepancies and resolves structural degradation through its unified framework.
  • The framework successfully utilizes edge cues, alleviating boundary issues and improving salient object detection accuracy.
  • Experimental results validate SECA-Net's effectiveness and superiority in salient object detection tasks.