Image Super-Resolution Using Very Deep Residual Channel Attention Networks
By Yulun Zhang, Kunpeng Li, Kai Li, Lichen Wang, Bineng Zhong, Yun Raymond Fu. 2018
5,405 citations recorded by the source corpus.
Convolutional neural network (CNN) depth is of crucial importance for image super-resolution (SR). However, we observe that deeper networks for image SR are more difficult to train. The low-resolution inputs and features contain abundant low-frequency information, which is treated equally across channels, hence hindering the representational ability of CNNs. To solve these problems, we propose the very deep residual channel attention networks (RCAN). Specifically, we propose a residual in residual (RIR) structure to form very deep network, which consists of several residual groups with long skip connections. Each residual group contains some residual blocks with short skip connections. Meanwhile, RIR allows abundant low-frequency information to be bypassed through multiple skip connections, making the main network focus on learning high-frequency information. Furthermore, we propose a channel attention mechanism to adaptively rescale channel-wise features by considering interdependencies among channels. Extensive experiments show that our RCAN achieves better accuracy and visual improvements against state-of-the-art methods.
Subfields: Computer Vision and Pattern Recognition, Media Technology
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Ming-Hsuan Yang
- Dacheng Tao
- Chen Change Loy
- Lei Zhang
- Radu Timofte
- Xiaogang Wang
- Yi Yang
In Media Technology
Citations in the corpus
- Image Super-Resolution via Iterative Refinement (2021, methodology): “Following [31], [64] we apply 4 super-resolution models to 56 56 center crops from the validation set of ImageNet.”
- Designing a Practical Degradation Model for Deep Blind Image Super-Resolution (2021, methodology): “As mentioned in the introduction, existing DNNs-based SISR methods are generally based on bicubic downsam-pling [23, 44] and traditional degradations [26, 37, 54, 59, 60], or some simple variants [11, 41, 53, 56, 58].”
- Hierarchical Neural Architecture Search for Single Image Super-Resolution (2020, methodology): “These two kinds of blocks/layers often follow a two-level architecture, where the network-level architecture determines the positions of the upsampling layers ( e.g. , SRCNN [7] and LapSRN [8]) and the cell-level architecture controls the computation of each block/layer ( e.g. , RCAB [9]).”
- Fast, Accurate and Lightweight Super-Resolution with Neural Architecture Search (2019, background): “Therefore, too deep and large models such as RDN [17], RCAN [18] are excluded here.”
- Selective Kernel Networks (2019, neutral): “Attention has been widely used in recent applications such as person re-ID [4], image recovery [55], text abstraction [34] and lip reading [48].”
- Pre-Trained Image Processing Transformer (2020, background): “[80] and Anwar and Barnes [3] utilize the power of attention to enhance the performance on SR task.”
- Restormer: Efficient Transformer for High-Resolution Image Restoration (2021, background): “In recent years, data-driven CNN architectures [7, 18, 92, 93, 105, 107] have been shown to outperform conventional restoration approaches [26,36,53,75].”
- Exploiting Diffusion Prior for Real-World Image Super-Resolution (2023, neutral): “Early SR approaches [9,13–16, 23, Recent works have moved their focus from synthetic settings to blind SR, where the degradation is unknown and similar to real-world scenarios.”
- Unsupervised Degradation Representation Learning for Blind Super-Resolution (2021, methodology): “Then, we compare our DASR to several recent SR networks, including RCAN [44], SRMD [42], MZSR [33] and IKC [12].”
- AIM 2020 Challenge on Real Image Super-Resolution: Methods and Results (2020, methodology): “The proposed method follows the main structure of RCAN [43] and utilizes the guild filter to decompose the detail layer and to restore high-frequency details.”
- Second-Order Attention Network for Single Image Super-Resolution (2019, methodology): “As in [38], we also add long and short skip connections in Base model.”
- Efficient Transformer for Single Image Super-Resolution (2021, methodology): “RCAN [18] proposes a very deep residual network with residual-in-residual architecture and channel attention mechanism.”
- EDVR: Video Restoration With Enhanced Deformable Convolutional Networks (2019, methodology): “Attention has proven its effectiveness in many tasks [43, 47, 22, 23, 51].”
- The 2018 PIRM Challenge on Perceptual Image Super-Resolution (2018, methodology): “While the first attempt to apply a conv-net to image SR [6] did not significantly surpass the performance of prior methods, it set the ground for major improvements in PSNR/SSIM values over the course of the several following years [15, 17, 18, 19, 39, 11, 52, 34, 10, 51].”