Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 89 of the 89 references it cites were found in a registry or an index.

Classification of Remote Sensing Images Using EfficientNet-B3 CNN Model With Attention

Open access
Alhichri, Haikel, Alswayed, Asma S., Bazi, Yakoub, Ammour, Nassim, Alajlan, Naif A. · 2021 · IEEE Access

Abstract

Scene classification is a highly useful task in Remote Sensing (RS) applications. Many efforts have been made to improve the accuracy of RS scene classification. Scene classification is a challenging problem, especially for large datasets with tens of thousands of images with a large number of classes and taken under different circumstances. One problem that is observed in scene classification is the fact that for a given scene, only one part of it indicates which class it belongs to, whereas the other parts are either irrelevant or they actually tend to belong to another class. To address this issue, this paper proposes a deep attention Convolutional Neural Network (CNN) for scene classification in remote sensing. CNN models use successive convolutional layers to learn feature maps from larger and larger regions (or receptive fields) of the scene. The attention mechanism computes a new feature map as a weighted average of these original feature maps. In particular, we propose a solution, named EfficientNet-B3-Attn-2, based on the pre-trained EfficientNet-B3 CNN enhanced with an attention mechanism. A dedicated branch is added to layer 262 of the network, to compute the required weights. These weights are learned automatically by training the whole CNN model end-to-end using the backpropagation algorithm. In this way, the network learns to emphasize important regions of the scene and suppress the regions that are irrelevant to the classification. We tested the proposed EfficientNet-B3-Attn-2 on six popular remote sensing datasets, namely UC Merced, KSA, OPTIMAL-31, RSSCN7, WHU-RS19, and AID datasets, showing its strong capabilities in classifying RS scenes.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1109/access.2021.3051085

Cites

89 of 89 references verified0 citation errors · 0 not checkable

Statistics

Figures

FIGURE 1 Sample RS scene from the ''storage tanks'' class. We can observe that some parts are irrelevant and belong to other classes (such as ''grass'').
FIGURE 1 Sample RS scene from the ''storage tanks'' class. We can observe that some parts are irrelevant and belong to other classes (such as ''grass'').CC-BY · From the PDF
FIGURE 3 The new Swish activation function [54]. Compared to ReU and LeakyReLU, the swish has a similar shape but is smoother.
FIGURE 3 The new Swish activation function [54]. Compared to ReU and LeakyReLU, the swish has a similar shape but is smoother.CC-BY · From the PDF
FIGURE 4 The EffecientNet-B0 general architecture.
FIGURE 4 The EffecientNet-B0 general architecture.CC-BY · From the PDF
FIGURE 5 Illustration of the squeeze-and-excitation concept [55].
FIGURE 5 Illustration of the squeeze-and-excitation concept [55].CC-BY · From the PDF
FIGURE 6 Illustration of MBConv6 with Squeeze-and-Excitation block and inverted residual connection 24 ! 144 ! 24.
FIGURE 6 Illustration of MBConv6 with Squeeze-and-Excitation block and inverted residual connection 24 ! 144 ! 24.CC-BY · From the PDF
FIGURE 7 Illustration of the MBConv1 block type.
FIGURE 7 Illustration of the MBConv1 block type.CC-BY · From the PDF
FIGURE 8 Illustration of the EffecientNet-B3 architecture. (10,3646 million weights). IRC means there is inverted residual connection.
FIGURE 8 Illustration of the EffecientNet-B3 architecture. (10,3646 million weights). IRC means there is inverted residual connection.CC-BY · From the PDF
FIGURE 9 Typical CNN architecture with two options after the last convolutional layer (a) The last feature vectors are flattened into one large vector, (b) The last feature vectors are averaged together.
FIGURE 9 Typical CNN architecture with two options after the last convolutional layer (a) The last feature vectors are flattened into one large vector, (b) The last feature vectors are averaged together.CC-BY · From the PDF
FIGURE 10 Details of the proposed attention module.
FIGURE 10 Details of the proposed attention module.CC-BY · From the PDF
FIGURE 11 EffecientNet-B3-Attn-2 (10,3646 million weights): attention module incorporated at some intermediate layer of the model.
FIGURE 11 EffecientNet-B3-Attn-2 (10,3646 million weights): attention module incorporated at some intermediate layer of the model.CC-BY · From the PDF
FIGURE 12 EffecientNet-B3-Attn-2: Attention mechanism is added to the original EfficientNet-B3 model as second branch connected to Block 19.
FIGURE 12 EffecientNet-B3-Attn-2: Attention mechanism is added to the original EfficientNet-B3 model as second branch connected to Block 19.CC-BY · From the PDF
FIGURE 13 Improved Images and their attention map for OPTIMAL-31 dataset with a 20%-80% train-test split using. Improvements from incorrect class to correct class after attention are (a) from Airport to Airplane, (b) from Basketball court to Playground, and (c) from Baseball field to Golf field.
FIGURE 13 Improved Images and their attention map for OPTIMAL-31 dataset with a 20%-80% train-test split using. Improvements from incorrect class to correct class after attention are (a) from Airport to Airplane, (b) from Basketball court to Playground, and (c) from Baseball field to Golf field.CC-BY · From the PDF
Image 13
CC-BY · From the PDF
Image 14
CC-BY · From the PDF
Image 15
CC-BY · From the PDF
Image 16
CC-BY · From the PDF
Image 17
CC-BY · From the PDF

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

1 retracted

Show all 86 references

Research works printed but not linked: 4

  1. alhichri, 2020, KSA Remote Sensing Datasets
  2. k?zrak, 2020, Comparison of activation functions for deep neural networks
  3. chen, 2018, Recurrent transformer networks for remote sensing scene categorisation, Proc Brit Mach Vis Conf
  4. castelluccio, 2015, Land use classification in remote sensing images by convolutional neural networks, arXiv 1508 00092

Cited by

See all 357

Metadata

Authors
Alhichri, H., Alswayed, A. S., Bazi, Y., Ammour, N., & Alajlan, N. A.
Affiliations
King Saud University (SA)
Venue
IEEE Access
Published
2021
Volume
9
Pages
14078-14094
ISSN
2169-3536 (ISSN-L)
OpenAlex
W3118660027
Cited by
357
References
86
Citations by year
2026: 40 · 2025: 81 · 2024: 83 · 2023: 68 · 2022: 60 · 2021: 25
FWCI
26.48
Type
journal article
Language
en
Open access
Locations
  • IEEE Access (publishedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
Researchers Supporting Project through the King Saud University, Riyadh, Saudi Arabia (RSP-2020/69)
Topics
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Remote-Sensing Image Classification
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
  • Social Sciences › Geography, Planning and Development › Remote Sensing and Land Use
Keywords
remote sensing scene classification, EfficientNetB3, attention mechanism, pre-trained CNN fine-tuning, feature map weighting

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Classification of Remote Sensing Images Using EfficientNet-B3 CNN Model With Attention. https://app.cochat.ai/papers/10.1109/access.2021.3051085
The paper
Alhichri, H., Alswayed, A. S., Bazi, Y., Ammour, N., & Alajlan, N. A. (2021). Classification of Remote Sensing Images Using EfficientNet-B3 CNN Model With Attention. IEEE Access. https://doi.org/10.1109/access.2021.3051085

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.