2021
DOI: 10.3390/rs13214411
|View full text |Cite
|
Sign up to set email alerts
|

Attention Enhanced U-Net for Building Extraction from Farmland Based on Google and WorldView-2 Remote Sensing Images

Abstract: High-resolution remote sensing images contain abundant building information and provide an important data source for extracting buildings, which is of great significance to farmland preservation. However, the types of ground features in farmland are complex, the buildings are scattered and may be obscured by clouds or vegetation, leading to problems such as a low extraction accuracy in the existing methods. In response to the above problems, this paper proposes a method of attention-enhanced U-Net for building… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
4
1

Citation Types

0
5
0

Year Published

2022
2022
2025
2025

Publication Types

Select...
8
1

Relationship

0
9

Authors

Journals

citations
Cited by 27 publications
(5 citation statements)
references
References 42 publications
0
5
0
Order By: Relevance
“…To further validate the effectiveness of the proposed method, the experimental results on two testing sets of TE-ResUNet are compared with existing methods in this section, including FCN8s [ 3 ], PSPNet [ 7 ], DeepLabv3 [ 8 ], and AEUNet [ 16 ]. Among these methods, FCN8s is a base semantic segmentation network with 8x upsampled prediction.…”
Section: Methodsmentioning
confidence: 99%
See 1 more Smart Citation
“…To further validate the effectiveness of the proposed method, the experimental results on two testing sets of TE-ResUNet are compared with existing methods in this section, including FCN8s [ 3 ], PSPNet [ 7 ], DeepLabv3 [ 8 ], and AEUNet [ 16 ]. Among these methods, FCN8s is a base semantic segmentation network with 8x upsampled prediction.…”
Section: Methodsmentioning
confidence: 99%
“…Compared with traditional machine learning algorithms, deep learning can extract high-level semantic features hidden in images, and has better performance in complex scenes. Therefore, deep learning based semantic segmentation methods open up a new way for remote sensing image earth object extraction [ 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 ].…”
Section: Introductionmentioning
confidence: 99%
“…Attention mechanisms in deep learning [ 36 , 37 ] are very similar to human visual attention mechanisms, which select more important information for the current target and remove redundant information. This allows the network to adaptively focus on the necessary information and can be achieved by using importance weight vectors to approximate the final target value through weighted vector summation.…”
Section: Introductionmentioning
confidence: 99%
“…This network structure does not perform well in the semantic segmentation of buildings, mainly due to the loss of the spatial information of local and regions as well as the poor, stride-convolved, and high-dimensional features. Therefore, some methods are proposed to compensate for local information [36,37] and improve the selection of the number of feature channels. Among them, PISANet model [28] obtains the global features and comprehensive features of the buildings through the pyramid self-attention module, which makes full use of the spatial information in the image; ARC-Net model [27] reduces the computational cost and shrinks the model size by residual blocks with asymmetric convolution, which improves the extraction effect based on reducing the model parameters.…”
Section: Introductionmentioning
confidence: 99%