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: Mar 20, 2026

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
04:23

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

Published on: April 21, 2023

2.4K

Multilayer pyramid pooling self-attention for landslide detection using vision transformers.

S Sreelakshmi1, S S Vinod Chandra2, Daud Ali3

  • 1Machine Intelligence Research Laboratory, Department of Computer Science, University of Kerala, Thiruvananthapuram, India. alimsdb@gmail.com.

Scientific Reports
|March 19, 2026
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

Advanced prognostic modeling with deep learning: assessing long-term outcomes in liver transplant recipients from deceased and living donors.

Journal of translational medicine·2025
Same author

Comparison: Imported and Indigenous HIV Fourth-generation immunodiagnostic kits.

Asian journal of transfusion science·2025
Same author

Recent progress in tannin and lignin blended metal oxides and metal sulfides as smart materials for electrochemical sensor applications.

Analytical sciences : the international journal of the Japan Society for Analytical Chemistry·2024
Same author

Lawsonia inermis-organically modified chitosan intercalated bentonite clay: A multifunctional nanotheranostic system for controlled drug delivery, sensing and cellular imaging.

International journal of biological macromolecules·2024
Same author

Significance of Sequence Features in Classification of Protein-Protein Interactions Using Machine Learning.

The protein journal·2023
Same author

Public discourse and sentiment during Mpox outbreak: an analysis using natural language processing.

Public health·2023

Vision Transformers (ViTs) are computationally expensive. This study integrates pyramid pooling into ViTs for efficient landslide detection, improving accuracy by preserving multi-scale context.

Area of Science:

  • Computer Vision
  • Remote Sensing
  • Geoscience

Background:

  • Vision Transformers (ViTs) excel in visual tasks but face computational challenges due to long sequences.
  • Sequence reduction via pooling is common, but single pooling methods limit contextual representation.
  • Pyramid pooling offers strong context abstraction capabilities.

Purpose of the Study:

  • To investigate the integration of pyramid pooling within a transformer framework for enhanced landslide detection in remote sensing imagery.
  • To adapt the Pyramid Pooling Transformer (PPT) for landslide analysis by incorporating a multi-layer pyramid pooling-based multi-head self-attention mechanism.
  • To enable efficient sequence reduction while preserving multi-scale contextual information for better spatial heterogeneity and scale variability capture.

Main Methods:

Keywords:
Computer visionDeep backbone networkLandslide detectionMachine learningPyramid poolingTransformer architectureVision transformer

Related Experiment Videos

Last Updated: Mar 20, 2026

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
04:23

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

Published on: April 21, 2023

2.4K
  • Building upon the existing Pyramid Pooling Transformer (PPT) backbone.
  • Incorporating a multi-layer pyramid pooling-based multi-head self-attention mechanism.
  • Task-oriented adaptation for landslide detection using remote sensing data.

Main Results:

  • The proposed PPT-based model significantly outperforms conventional Convolutional Neural Networks (CNNs) and standard transformer baselines.
  • Achieved a 7.3% improvement in F1-score and a 2% increase in overall accuracy compared to state-of-the-art deep learning models.
  • Demonstrated effective capture of spatial heterogeneity and scale variability in landslide scenes.

Conclusions:

  • Pyramid pooling-driven attention mechanisms are highly effective for landslide detection.
  • The adapted PPT model offers a computationally efficient and accurate solution for remote sensing-based landslide analysis.
  • This approach enhances the performance of transformers in specialized remote sensing applications.