Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Video

Updated: May 28, 2026

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments

Published on: December 15, 2023

BATFNet: Boundary-Aware Transformer Fusion Network for RGB-DSM Semantic Segmentation of Remote Sensing Images.

Yilin Tong¹, Meng Tang², Yu Zhang³

¹School of Electronic Engineering, Wuhan Vocational College of Software and Engineering, Wuhan 430205, China.

Sensors (Basel, Switzerland)

|May 27, 2026

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

Cu(II)-Catalyzed Cascade Cyclization and Oxidative Annulation to Access Functionalized 1<i>H</i>-Furo[3,2-<i>b</i>]pyrazolo[4,3-<i>e</i>]pyridines.

Organic letters·2026

Same author

Application of a glyceride-mimetic prodrug strategy to deliver protease inhibitors to gut lymph to alleviate organ failure.

Journal of controlled release : official journal of the Controlled Release Society·2026

Same author

From Laboratory to Clinic: Translational Medicine Paradigm of Polymyxin B Nanopreparations for Overcoming Drug-Resistant Bacterial Infections.

International journal of nanomedicine·2026

Same author

Influence of High-Volume Calcined Phosphogypsum on Mechanical Properties and Freeze-Thaw Resistance of Supersulfated Slag Cement Concrete.

Materials (Basel, Switzerland)·2026

Same author

Thermally synergized electron shuttling via Cu-O-Fe bridge redirects H<sub>2</sub>O<sub>2</sub> activation pathways.

Water research·2026

Same author

A web-based platform for real-time stewed beef freshness monitoring: Integrating anthocyanin colorimetric film with deep learning.

Food chemistry·2026

Combining RGB appearance and Digital Surface Model (DSM) height data improves semantic segmentation in remote sensing. A new boundary-aware Transformer fusion network (BATFNet) enhances urban land-cover classification accuracy.

Area of Science:

Remote Sensing
Computer Vision
Geospatial Analysis

Background:

Semantic segmentation of remote sensing imagery is crucial for urban analysis.
Integrating RGB appearance and Digital Surface Model (DSM) height data offers complementary information.
Urban scenes present challenges due to spectrally similar objects with varying elevations.

Purpose of the Study:

To develop and evaluate a novel fusion network for semantic segmentation of very-high-resolution remote sensing imagery.
To leverage both spectral (RGB) and structural (DSM) information for improved classification accuracy.
To enhance boundary delineation and reduce misclassification of spectrally similar urban land-cover categories.

Main Methods:

Proposed BATFNet: a supervised boundary-aware Transformer fusion network.

Keywords:

cross-modal attention deep learning remote sensing semantic segmentation

Related Experiment Videos

Last Updated: May 28, 2026

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments

Published on: December 15, 2023

Utilized a dual-branch ResNet-50 backbone for modality-specific feature extraction.

Employed DSM-derived edge priors to guide bidirectional cross-modal attention and decoder refinement.

Main Results:

Achieved mIoU scores of 84.06% (Vaihingen) and 85.31% (Potsdam) on benchmark datasets.
Outperformed representative RGB-DSM fusion baselines across most land-cover categories.
Demonstrated effective integration of RGB and DSM data, recovering fine spatial details.

Conclusions:

Exploiting DSM-derived structural cues significantly improves boundary delineation in semantic segmentation.
The proposed BATFNet effectively reduces confusion among spectrally similar urban classes.
Combining appearance and height information is highly beneficial for remote sensing semantic segmentation, particularly in complex urban environments.