LatticeNet: Towards Lightweight Image Super-Resolution with Lattice Block

Luo, Xiaotong; Xie, Yuan; Zhang, Yulun; Qu, Yanyun; Li, Cuihua; Fu, Yun

doi:10.1007/978-3-030-58542-6_17

Cited by 174 publications

(83 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recently, some fast and lightweight SISR architectures have been introduced to tackle image SR. These methods can be approximately divided into three categories: the knowledge distillation-based methods [19,27,28], the neural architecture search-based methods [41,42], and the model design-based methods [26,43]. Knowledge distillation aims to transfer the knowledge from a teacher network to a student network.…”

Section: Single Image Super-resolutionmentioning

confidence: 99%

“…In addition, CARN [26] proposes a cascading mechanism based on a residual network to boost performance. LatticeNet [43] proposes a lattice block in which two butterfly structures are applied to combine two residual blocks. These works indicate that the lightweight SR networks can maintain a good trade-off between performance and model complexity.…”

Section: Single Image Super-resolutionmentioning

confidence: 99%

See 1 more Smart Citation

URNet: A U-Shaped Residual Network for Lightweight Image Super-Resolution

et al. 2021

View full text Add to dashboard Cite

It is extremely important and necessary for low computing power or portable devices to design more lightweight algorithms for image super-resolution (SR). Recently, most SR methods have achieved outstanding performance by sacrificing computational cost and memory storage, or vice versa. To address this problem, we introduce a lightweight U-shaped residual network (URNet) for fast and accurate image SR. Specifically, we propose a more effective feature distillation pyramid residual group (FDPRG) to extract features from low-resolution images. The FDPRG can effectively reuse the learned features with dense shortcuts and capture multi-scale information with a cascaded feature pyramid block. Based on the U-shaped structure, we utilize a step-by-step fusion strategy to improve the performance of feature fusion of different blocks. This strategy is different from the general SR methods which only use a single Concat operation to fuse the features of all basic blocks. Moreover, a lightweight asymmetric residual non-local block is proposed to model the global context information and further improve the performance of SR. Finally, a high-frequency loss function is designed to alleviate smoothing image details caused by pixel-wise loss. Simultaneously, the proposed modules and high-frequency loss function can be easily plugged into multiple mature architectures to improve the performance of SR. Extensive experiments on multiple natural image datasets and remote sensing image datasets show the URNet achieves a better trade-off between image SR performance and model complexity against other state-of-the-art SR methods.

show abstract

Section: Single Image Super-resolutionmentioning

confidence: 99%

Section: Single Image Super-resolutionmentioning

confidence: 99%

URNet: A U-Shaped Residual Network for Lightweight Image Super-Resolution

et al. 2021

View full text Add to dashboard Cite

show abstract

“…Recently, some researchers have proposed attention-based models to improve the SR performance. Several works, such as RCAN [6], SAN [21], MCAN [47], A 2 F [48], LatticeNet [49], and DIN [50], introduce the channel attention (CA) mechanism to SR which makes the network learn more useful features. To learn more discriminative features, some researchers utilize both the channel attention and spatial attention, such as HRAN [51], MIRNet [52], CSFM [53], and BAM [54].…”

Section: Attention Mechanismmentioning

confidence: 99%

Residual Triplet Attention Network for Single-Image Super-Resolution

Huang

Wang

et al. 2021

Electronics

View full text Add to dashboard Cite

Single-image super-resolution (SISR) techniques have been developed rapidly with the remarkable progress of convolutional neural networks (CNNs). The previous CNNs-based SISR techniques mainly focus on the network design while ignoring the interactions and interdependencies between different dimensions of the features in the middle layers, consequently hindering the powerful learning ability of CNNs. In order to address this problem effectively, a residual triplet attention network (RTAN) for efficient interactions of the feature information is proposed. Specifically, we develop an innovative multiple-nested residual group (MNRG) structure to improve the learning ability for extracting the high-frequency information and train a deeper and more stable network. Furthermore, we present a novel lightweight residual triplet attention module (RTAM) to obtain the cross-dimensional attention weights of the features. The RTAM combines two cross-dimensional interaction blocks (CDIBs) and one spatial attention block (SAB) base on the residual module. Therefore, the RTAM is not only capable of capturing the cross-dimensional interactions and interdependencies of the features, but also utilizing the spatial information of the features. The simulation results and analysis show the superiority of the proposed RTAN over the state-of-the-art SISR networks in terms of both evaluation metrics and visual results.

show abstract

“…At present, the excellent lightweight SR network models are LatticeNet [27], SLUA [28], MAFFSRN [29], and RFDN [30]. The MAFFSRN network initially uses a convolution to extract features and then uses many FFGs to refine and enhance features.…”

Section: Related Research Backgroundmentioning

confidence: 99%

Light Weight IBP Deep Residual Network for Image Super Resolution

Lin

Yang

2021

IEEE Access

View full text Add to dashboard Cite

Single -image super resolution (SR) is used to reconstruct a high-resolution image with more high-frequency details based on a low-resolution image as input. In recent years, image SR reconstruction based on deep learning methods has shown a considerably better performance than traditional methods. Early deep-learning-based methods deepen convolutional layers and directly reconstruct high-resolution images with complex neural networks. However, with the stacking of modules, network depth and model parameters increase, thereby raising computational resource; hence, it is difficult to apply on lowconfiguration devices. Furthermore, existing methods ignore the high-frequency details of the image, resulting in unsatisfactory performance. To solve these problems, a lightweight network model that applies the iterative back projection (IBP) mechanism to the network and reduces the dimensionality of the input image features is proposed. The proposed network model consists of three parts, namely, entrance module, main body module, and exit module. It designs a lightweight modular design to reduce the model calculation and control the network depth more easily by adjusting the number of ADB modules. The main part of the model consists of four lightweight accelerating deep residual back projection (ADB) modules. Each ADB module initially decreases the dimensionality of the input image features through a 1×1 convolutional layer to reduce the amount of calculation. Then, IBP is used to back project the image iteratively. Each ADB module only performs the downsampling back projection operation because the image features become larger after upsampling. Through three iterative downsampling back projection units, the high-frequency features of the image are fully explored, and the output image features are then compared with the shallow layer. Image feature fusion is used as input of the next ADB module, and the output part combines the highand low-frequency image feature output by multiple ADB modules to complete the upsampling by using the PixelShuffle method to generate high-resolution images. Several experiments confirm that the proposed algorithm achieves better SR reconstruction accuracy with faster reconstruction speed than existing image SR methods. INDEX TERMS image SR; residual network; light-weight convolutional neural network;

show abstract

LatticeNet: Towards Lightweight Image Super-Resolution with Lattice Block

Cited by 174 publications

References 31 publications

URNet: A U-Shaped Residual Network for Lightweight Image Super-Resolution

URNet: A U-Shaped Residual Network for Lightweight Image Super-Resolution

Residual Triplet Attention Network for Single-Image Super-Resolution

Light Weight IBP Deep Residual Network for Image Super Resolution

Contact Info

Product

Resources

About