Improving Super-Resolution Performance Using Meta-Attention Layers

Aquilina, Matthew; Galea, Christian; Abela, John; Camilleri, Kenneth P.; Farrugia, Reuben A.

doi:10.1109/lsp.2021.3116518

Cited by 10 publications

(9 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Joinput-CiNet [2]: It is a SCI network with joint input of degradation maps and LR measurements, which uses PCA (principal component analysis) to extract sensing matrix information to guide reconstruction. Meta-CiNet [54], [55]: It is an improved version of Joinput-CiNet, which extracts more dimensional information of the sensing matrix than the former. RCAN [53]:This is one of the most representative CNN SR networks, which uses residual-in-residual and channel attention mechanism to build a very deep network to achieve high-quality reconstruction.…”

Section: B Comparing Sci Methodsmentioning

confidence: 99%

Meta-TR: Meta-Attention Spatial Compressive Imaging Network With Swin Transformer

Cui

Yang

et al. 2022

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

As a flourishing research topic in the field of remote sensing, SCI (spatial compressive imaging) can utilize prior knowledge to recover high-dimensional signals from LR (lowresolution) measurements through joint sampling and compression, thus contributing to the bandwidth reduction of information transmission. However, most of the existing SCI methods based on deep learning cannot effectively utilize prior information, and difficult to perform deep extraction of image features, so the reconstruction is not ideal in the case of low sampling ratio. To address the above difficulty, we propose a SCI network based on MA (meta-attention) and swin transformer, named Meta-TR. We adopt the swin transformer as the network backbone, through the wide application of self-attention mechanisms, to achieve deeper extraction of image features, thereby improving the reconstruction quality under low sampling ratios. In addition, we design a meta-attention module, which adopts Squeeze-Excitation architecture to convert the metadata of SCI image degradation process to attention vectors. Then the attention vectors are used in the channel modulation of network feature maps to guide the network training. Extensive experiments are performed on different benchmark remote sensing datasets and different sampling ratios to confirm the superiority of the proposed Meta-TR method.

show abstract

Section: B Comparing Sci Methodsmentioning

confidence: 99%

Meta-TR: Meta-Attention Spatial Compressive Imaging Network With Swin Transformer

Cui

Yang

et al. 2022

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

show abstract

“…Each of these approaches showed that SR networks could make use of this degradation information, improving their SR performance as a result. Frameworks to enable the extension of existing non-blind SR methods to use degradation information have also been proposed, such as the 'meta-attention' approach in [25] and Conditional hyper-network framework for SR with Multiple Degradations (CMDSR) proposed in [55].…”

Section: Approaches Utilising Supplementary Attributes For Srmentioning

confidence: 99%

“…In our work, we extended this methodology to all degradation vectors considered. • MA [25]: MA is a trainable channel attention block which was proposed as a way to upgrade any CNN-based SR network with metadata information. Its functionality is simple -an input vector is stretched to the same size as the number of feature maps within a target CNN network using two fully-connected layers.…”

Section: Metadata Insertion Blockmentioning

confidence: 99%

“…We implement a variety of SR architectures and integrate these with the latest techniques for contrastive learning and iterative degradation prediction. Our results show that by using just a single Meta-Attention (MA) layer [25], high-performance SR models such as the Residual Channel Attention Network (RCAN) [7] and Holistic Attention Network (HAN) [9] can be infused with degradation information to yield SR results which outperform those of the original blind SR networks trained under the same conditions.…”

Section: Introductionmentioning

confidence: 99%

“…For a full exposition on the selection of each type of kernel, please refer to [15] [30,95] (following [15]). For JM H.264, images were compressed as single-frame YUV files where a random I-slice Quantization Parameter (QPI) were selected from the range [20,40], following [25]. All our models were trained on the LR images generated from the HR images of DIV2K [78] (800 images) and Flickr2K [79] (2,650 images).…”

mentioning

confidence: 99%

See 2 more Smart Citations