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: Dec 9, 2025

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

875

Spatial-Spectral Feature Refinement for Hyperspectral Image Classification Based on Attention-Dense 3D-2D-CNN.

Jin Zhang1, Fengyuan Wei1, Fan Feng1

  • 1School of Surveying and Land Information Engineering, Henan Polytechnic University, Jiaozuo 454003, China.

Sensors (Basel, Switzerland)
|September 16, 2020
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

Involvement of COX-2 in the Neurovascular Unit Damage Through Up-Regulating Calpain/PARP/NF-κB Inflammatory Signaling During Ischemic Stroke.

Neurochemical research·2026
Same author

Trends and health inequalities in the global burden of kidney cancer from 1990 to 2021, with projections to 2040.

Medicine·2026
Same author

Gapless comparative genomics reveals effector landscape and adaptive selection signatures in Plasmodiophora brassicae.

Journal of genetics and genomics = Yi chuan xue bao·2026
Same author

Atomic Origins of Ultrahigh-Voltage Failure in LiCoO<sub>2</sub> Cathodes.

Journal of the American Chemical Society·2026
Same author

Multisite Chronic Pain Reveals Neuro-Immune-Metabolic Dysregulation across Rheumatoid Arthritis and Depression.

Research (Washington, D.C.)·2026
Same author

The Distally Based Saphenous Neurocutaneous Perforator Flap: A Versatile Donor Site for Reconstruction of Complex Soft Tissue Defects of the Lower Leg.

Annals of plastic surgery·2026

This study introduces the Attention-Dense-HybridSN (AD-HybridSN) model for hyperspectral image (HSI) classification with limited training data. The novel approach significantly improves classification accuracy even with minimal labeled samples.

Area of Science:

  • Remote Sensing
  • Computer Vision
  • Machine Learning

Background:

  • Hyperspectral image (HSI) classification is crucial for various applications.
  • Limited training samples pose a significant challenge for existing deep learning models.
  • Convolutional Neural Networks (CNNs) show promise but require sufficient data.

Purpose of the Study:

  • To develop a novel CNN model for effective HSI classification under limited training data conditions.
  • To enhance the exploitation of spatial-spectral features in HSI data.
  • To address the
  • small sample
  • problem in HSI classification.

Main Methods:

  • Proposed a novel 3D-2D-convolutional neural network (CNN) named AD-HybridSN (Attention-Dense-HybridSN).
Keywords:
3D-2D-CNNattention mechanismdeep learninghyperspectral image classificationresidual connectionspatial–spectral feature refinement

More Related Videos

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
04:48

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Published on: July 5, 2024

676

Related Experiment Videos

Last Updated: Dec 9, 2025

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

875
Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
04:48

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Published on: July 5, 2024

676
  • Utilized dense blocks for shallow feature reuse and hierarchical spatial-spectral feature extraction.
  • Employed depth separable convolutional layers for spatial information discrimination.
  • Integrated channel and spatial attention mechanisms to refine features after 3D and 2D convolutional layers.

    • Main Results:

    • AD-HybridSN demonstrated superior performance in hyperspectral image classification with very few training samples.
    • Achieved high overall accuracies: 97.02% on Indian Pines, 99.59% on Salinas, and 98.32% on the University of Pavia dataset.
    • These results were obtained using only 5%, 1%, and 1% labeled data for training, respectively.
    • Outperformed all contrast models in small-sample HSI classification tasks.

    Conclusions:

    • The AD-HybridSN model effectively learns discriminative spatial-spectral features from limited hyperspectral data.
    • The proposed attention-dense hybrid network architecture is highly effective for small-sample HSI classification.
    • This work offers a significant advancement for HSI classification in data-scarce scenarios.