2021
DOI: 10.1109/tcsvt.2020.3028008
|View full text |Cite
|
Sign up to set email alerts
|

Multi-Deformation Aware Attention Learning for Concrete Structural Defect Classification

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
2

Citation Types

0
12
0

Year Published

2021
2021
2024
2024

Publication Types

Select...
7
1

Relationship

0
8

Authors

Journals

citations
Cited by 20 publications
(12 citation statements)
references
References 24 publications
0
12
0
Order By: Relevance
“…Use of visual attention in concrete defect classification is however limited; except a recent work [47] involving residual attention mechanism. However, the considerable amount of parameters and computations involved in parallel feature extraction make the network inferior for real-time applications.…”
Section: Attention Mechanismmentioning
confidence: 99%
See 2 more Smart Citations
“…Use of visual attention in concrete defect classification is however limited; except a recent work [47] involving residual attention mechanism. However, the considerable amount of parameters and computations involved in parallel feature extraction make the network inferior for real-time applications.…”
Section: Attention Mechanismmentioning
confidence: 99%
“…Similarly, we perform spatial squeezing operation inside simultaneous excitation module using 1 × 1 convolution, rather than using the pooling operation in [22], [23] to adaptively generate local spatial descriptors. Moreover, the iDAAM architecture attributes to significant reduction in the number of parameters compared to the state-of-the-art [47] due to the absence of explicit multi-branch feature extraction.…”
Section: Attention Mechanismmentioning
confidence: 99%
See 1 more Smart Citation
“…Thus, it is important to construct a model that takes into account the practical inspection process, and various studies have been conducted focusing on the above first point [13][14][15][16][17]. For example, in the literature [16], the global average pooling layer is inserted into the originally constructed model, and the class activation map proposed in the field of general object recognition [18][19][20][21] is used to reveal the regions that the model paid attention to during the estimation.…”
Section: Introductionmentioning
confidence: 99%
“…The presented method used five specific parameters, concrete samples, and acquired images. Multitarget multiclass defects in concrete structures found in civil infrastructures were classified using a deep multi-deformation-aware attention learning architecture, comprising a multiscale assembly of attention and fine-grained feature-induced attention modules [11]. An image-processing technique to provide concrete regression analysis has been employed [12]; the technique can be an auxiliary tool for destructive and nondestructive testing methods.…”
Section: Introductionmentioning
confidence: 99%