Residual Triplet Attention Network for Single-Image Super-Resolution

Huang, Feng; Wang, Zhifeng; Wu, Jing; Shen, Ying; Chen, Liqiong

doi:10.3390/electronics10172072

Cited by 3 publications

(4 citation statements)

References 63 publications

(114 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Hui et al [15] designed the information multi-distillation network (IMDN) for better purifying each processing feature by explicitly splitting the preceding features into two segments, whereby one part is maintained and the other is further refined, streamlining the network parameters and boosting the reconstruction performance. RTAN [21] is a novel lightweight residual triplet attention module used to obtain the cross-dimensional attention weights of the features. LW-AWSRN [22] is a novel local fusion block consisting of adaptive weighted residual units and the local residual fusion unit to remove the redundancy scale branch.…”

Section: Lightweight Cnns For Srmentioning

confidence: 99%

Infrared Image Super-Resolution via Progressive Compact Distillation Network

Fan

Hong

2021

Electronics

View full text Add to dashboard Cite

Deep convolutional neural networks are capable of achieving remarkable performance in single-image super-resolution (SISR). However, due to the weak availability of infrared images, heavy network architectures for insufficient infrared images are confronted by excessive parameters and computational complexity. To address these issues, we propose a lightweight progressive compact distillation network (PCDN) with a transfer learning strategy to achieve infrared image super-resolution reconstruction with a few samples. We design a progressive feature residual distillation (PFDB) block to efficiently refine hierarchical features, and parallel dilation convolutions are utilized to expand PFDB’s receptive field, thereby maximizing the characterization power of marginal features and minimizing the network parameters. Moreover, the bil-global connection mechanism and the difference calculation algorithm between two adjacent PFDBs are proposed to accelerate the network convergence and extract the high-frequency information, respectively. Furthermore, we introduce transfer learning to fine-tune network weights with few-shot infrared images to obtain infrared image mapping information. Experimental results suggest the effectiveness and superiority of the proposed framework with low computational load in infrared image super-resolution. Notably, our PCDN outperforms existing methods on two public datasets for both ×2 and ×4 with parameters less than 240 k, proving its efficient and excellent reconstruction performance.

show abstract

Section: Lightweight Cnns For Srmentioning

confidence: 99%

Infrared Image Super-Resolution via Progressive Compact Distillation Network

Fan

Hong

2021

Electronics

View full text Add to dashboard Cite

show abstract

“…The direct data processing technique involves immediately applying the graph convolution analysis approach to the 3D point cloud data without first subjecting it to voxel filtering or multi-view conversion. The two most crucial deep learning techniques are PointNet [11], PointNet++ [12], and their enhanced algorithms that followed [13][14][15][16][17][18][19][20].…”

Section: Introductionmentioning

confidence: 99%

“…For the disorganized, unstructured, densely packed but sparsely distributed properties of point cloud data. The T-net method, point-by-point multi-layer perceptron, and channelby-channel maximum pooling method are used by PointNet [11] to learn the general properties of 3D point cloud data and to address the issue of disorder in this data. The algorithm itself has the flaw of being unable to determine what the local attributes of the point cloud are.…”

Section: Introductionmentioning

confidence: 99%

Graph Convolution Algorithm Based on Visual Selectivity and Point Cloud Analysis Application

Zhao¹,

Su²,

Ye³

et al. 2022

View full text Add to dashboard Cite

The graph convolution algorithm currently suffers from the drawback of not fusing point cloud information and point cloud topology structure information based on visual selectivity features and using absolute quantities like distance as features, resulting in the algorithm losing geometric invariance. This information serves as the foundation for the "Graph Convolution Algorithm Based on Visual Selectivity and Application of Point Cloud Analysis". In order to propose a graph convolutional kernel and its design method based on visual selectivity, the algorithm analyzes the global characteristics of the point cloud "close in the vicinity and sparse in the distance," the local selectivity of the point cloud topology structure in the neighborhood, and the consistency between features and visual selectivity of primates. By combining point cloud information with point cloud topology structure information features, a graph convolution computation method was built, and the algorithm's geometric invariance was confirmed. The recognition and semantic segmentation performances of the approach in this study were verified using the ModelNet40 and ShapeNetPart data sets in comparison to the PointNet, PointNet++, DGCNN, KPConv, and 3D-GCN algorithms. The experimental design demonstrates that the algorithm presented in this research is accurate and practical, has geometric invariance, and performs better at semantic segmentation and recognition than conventional algorithms.

show abstract

“…Method [19] proposes a two-stage multi-task network to focus on image contrast and local brightness during network training. Method [20] proposed a threelayer residual attention network to solve the interaction problem of different dimensional features in the middle layers of the network during training.…”

Section: Introductionmentioning

confidence: 99%

High-Magnification Super-Resolution Reconstruction of Image with Multi-Task Learning

Zhu

2022

Electronics

View full text Add to dashboard Cite

Single-image super-resolution technology has made great progress with the development of the convolutional neural network, but most of the current super-resolution methods do not attempt high-magnification image super-resolution reconstruction; only reconstruction with ×2, ×3, ×4 magnification is carried out for low-magnification down-sampled images without serious degradation. Based on this, this paper proposed a single-image high-magnification super-resolution method, which extends the scale factor of image super-resolution to high magnification. By introducing the idea of multi-task learning, the process of the high-magnification image super-resolution process is decomposed into different super-resolution tasks. Different tasks are trained with different data, and network models for different tasks can be obtained. Through the cascade reconstruction of different task network models, a low-resolution image accumulates reconstruction advantages layer by layer, and we obtain the final high-magnification super-resolution reconstruction results. The proposed method shows better performance in quantitative and qualitative comparison on the benchmark dataset than other super-resolution methods.

show abstract

Residual Triplet Attention Network for Single-Image Super-Resolution

Cited by 3 publications

References 63 publications

Infrared Image Super-Resolution via Progressive Compact Distillation Network

Infrared Image Super-Resolution via Progressive Compact Distillation Network

Graph Convolution Algorithm Based on Visual Selectivity and Point Cloud Analysis Application

High-Magnification Super-Resolution Reconstruction of Image with Multi-Task Learning

Contact Info

Product

Resources

About