Deeply-Recursive Convolutional Network for Image Super-Resolution

By Jiwon Kim, Jung Kwon Lee, Kyoung Mu Lee. 2015

2,772 citations recorded by the source corpus.

We propose an image super-resolution method (SR) using a deeply-recursive convolutional network (DRCN). Our network has a very deep recursive layer (up to 16 recursions). Increasing recursion depth can improve performance without introducing new parameters for additional convolutions. Albeit advantages, learning a DRCN is very hard with a standard gradient descent method due to exploding/ vanishing gradients. To ease the difficulty of training, we propose two extensions: recursive-supervision and skip-connection. Our method outperforms previous methods by a large margin.

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

In Media Technology

Citations in the corpus

Image Super-Resolution via Iterative Refinement (2021, neutral): “Numerous super-resolution methods have been proposed [16], [30], [31], [32], [49], [50], [51], [52], [53].”
Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network (2016, methodology): “We validate the proposed approaches using images from publicly available benchmark datasets and compare our performance against state-of-the-art works including SRCNN [8], SelfExSR [24] and DRCN [27].”
Why Should We Add Early Exits to Neural Networks? (2020, methodology): “A variant of the JT approach is described in [31].”
Hierarchical Neural Architecture Search for Single Image Super-Resolution (2020, methodology): “In this experiment, we compare our method with several hand-crafted SR models, including the Bicubic, SRCNN [7], VDSR [27], DRCN [28], DRRN [29], SelNet [30], CARN [31].”
Fast, Accurate and Lightweight Super-Resolution with Neural Architecture Search (2019, methodology): “In addition, it dominates DRCN [Kim et al. , 2016b] and MoreMNAS-A [Chu et al. , 2019] over three objectives on four datasets.”
EnhanceNet: Single Image Super-Resolution Through Automated Texture Synthesis (2016, background): “Figure 5 gives an overview of different approaches including the current state of the art by PSNR [25, 26] on the zebra image from Set14 which is particularly well-suited for a visual comparison since it contains both smooth and sharp edges, textured regions as well as repeating patterns.”
ESRGAN: Enhanced Super-Resolution Generative Adversarial Networks (2018, background): “orks. Later on, the eld has witnessed a variety of network architectures, such as a deeper network with residual learning [5], Laplacian pyramid structure [6], residual blocks [1], recursive learning [7,8], densely connected network [9], deep back projection [10] and residual dense network [11]. Spe”
Palette: Image-to-Image Diffusion Models (2021, background): “, multi-scale super-resolution [Kim et al. 2016], and JPEG restoration on multiple quality factors [Galteri et al.”
Enhanced Deep Residual Networks for Single Image Super-Resolution (2017, background): “Recently, deep neural networks [11, 12, 14] provide significantly improved performance in terms of peak signal-tonoise ratio (PSNR) in the SR problem.”
Deep Laplacian Pyramid Networks for Fast and Accurate Super-Resolution (2017, methodology): “, SRCNN [7], SCN [33], VDSR [17], and DRCN [18].”
Image Super-Resolution Using Very Deep Residual Channel Attention Networks (2018, methodology): “Evaluation Bicubic DRCN FSRCNN PSyCo ENet-E RCAN Baseline [19] [2] [45] [8] (ours)”
Efficient Transformer for Single Image Super-Resolution (2021, methodology): “RNN-based models [1], [2], [3], [4] decrease the number of parameters effectively compared to the standard CNN and obtain good performance.”
Residual Dense Network for Image Super-Resolution (2018, methodology): “increased the network depth in VDSR [10] and DRCN [11] by using gradient clipping, skip connection, or recursive-supervision to ease the difficulty of training deep network.”
Correction to: Single Image Super-Resolution via a Holistic Attention Network (2020, methodology): “To alleviate the phenomenon of gradient explosion and reduce the complexity of the model, DRCN [16] and DRRN [30] were proposed by using a recursive convolutional network.”
Neural Proximal Gradient Descent for Compressive Imaging (2018, background): “Alternatively, to facilitate the training one can ask reconstructions at each iteration to be faithful with the ground-truth images as in [18].”
Efficient Residual Dense Block Search for Image Super-Resolution (2019, methodology): “Recursive operator has been widely investigated to reduce redundant parameters of super-resolution network, such as DRCN (Kim, Kwon Lee, and Mu Lee 2016b), DRRN (Tai, Yang, and Liu 2017), and MemNet (Tai et al. 2017).”
Wide Activation for Efficient and Accurate Image Super-Resolution (2018, methodology): “To address these issues, recursive neural networks [15, 32] are proposed by re-using weights repeatedly.”
Image Super-Resolution by Neural Texture Transfer (2019, methodology): “741 DRCN [22] 25.”
Image Super-Resolution via Progressive Cascading Residual Network (2018, methodology): “One may meet this need by combining the recursive network [14, 18] with our method so that the parameters of the modules in all stages become tied.”
Dual Reconstruction Nets for Image Super-Resolution with Gradient Sensitive Loss (2018, methodology): “We compare the performance of our proposed DRN approach with several state-of-the-art methods, including Bicubic, SRCNN [9], SelfExSR [17], DRCN [22], ESPCN [38], SRResNet [25], SRGAN [25], FSRCNN [10], VDSR [21], DRRN [40], LapSRN [24], SRDenseNet [42] and EDSR [28].”