Unsupervised Real-world Image Super Resolution via Domain-distance Aware Training

@article{Wei2021UnsupervisedRI,
  title={Unsupervised Real-world Image Super Resolution via Domain-distance Aware Training},
  author={Yunxuan Wei and Shuhang Gu and Yawei Li and Longcun Jin},
  journal={2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year={2021},
  pages={13380-13389}
}
These days, unsupervised super-resolution (SR) is soaring due to its practical and promising potential in real scenarios. The philosophy of off-the-shelf approaches lies in the augmentation of unpaired data, i.e. first generating synthetic low-resolution (LR) images ${\mathcal{Y}^g}$ corresponding to real-world high-resolution (HR) images ${\mathcal{X}^r}$ in the real-world LR domain ${\mathcal{Y}^r}$, and then utilizing the pseudo pairs $\left\{ {{\mathcal{Y}^g},{\mathcal{X}^r}} \right\}$ for… 

Figures and Tables from this paper

Unsupervised Real-World Super-Resolution: A Domain Adaptation Perspective

TLDR
A novel unpaired SR training framework based on feature distribution alignment, with which one can obtain degradation-indistinguishable feature maps and then map them to HR images and indicates that this network obtains the state-of-the-art performance over both blind and unpairedSR methods on diverse datasets.

Real-World Image Super-Resolution by Exclusionary Dual-Learning

TLDR
A method, Real- World image Super-Resolution by Exclusionary Dual-Learning (RWSR-EDL) to address the feature diversity in perceptual- and L1- based cooperative learning and a noise-guidance data collection strategy is developed to addressed the training time con- sumption in multiple datasets optimization.

Unpaired Image Super-Resolution with Optimal Transport Maps

TLDR
An algorithm is proposed for unpaired SR which learns an unbiased OT map for the perceptual transport cost and provides nearly state-of-the-art performance on the large-scale unpaired AIM-19 dataset.

Frequency Consistent Adaptation for Real World Super Resolution

TLDR
A novel Frequency Consistent Adaptation (FCA) that ensures the frequency domain consistency when applying existing Super-Resolution methods to the real scene and improves the performance of the SR model under real-world setting achieving state-of-the-art results with high fidelity and plausible perception.

Self-Supervised Learning for Real-World Super-Resolution from Dual Zoomed Observations

TLDR
This paper presents a novel self-supervised learning approach for real-world image SR from observations at dual camera zooms (SelfDZSR), and takes the telephoto image instead of an additional high-resolution image as the supervision information and selects a center patch from it as the reference to super-resolve the corresponding short-focus image patch.

Learning to Zoom-In via Learning to Zoom-Out: Real-World Super-Resolution by Generating and Adapting Degradation

TLDR
A framework to learn SR from an arbitrary set of unpaired LR and HR images and to see how far a step can go in such a realistic and “unsupervised” setting is proposed.

Criteria Comparative Learning for Real-scene Image Super-Resolution

TLDR
This work proposes a novel training paradigm for RealSR, named Criteria Comparative Learning (Cria-CL), by developing contrastive losses on criteria instead of image patches, and demonstrates that compared with the typical weighted regression strategy, this method achieves a significant improvement under similar parameter settings.

Unpaired Real-World Super-Resolution with Pseudo Controllable Restoration

TLDR
This paper proposes an unpaired real-world super-resolution method that performs on par, or even better than blind paired approaches by introducing a pseudo-controllable restoration module in a fully end-to-end system.

Kernel Adversarial Learning for Real-world Image Super-resolution

TLDR
A more realistic process to lower image resolution is proposed by introducing a new Kernel Adversarial Learning Super-resolution (KASR) framework to deal with the real-world image SR problem.

Hierarchical Conditional Flow: A Unified Framework for Image Super-Resolution and Image Rescaling

TLDR
The hierarchical conditional flow (HCFlow) is proposed as a unified framework for image SR and image rescaling and achieves state-of-the-art performance in terms of both quantitative metrics and visual quality.

References

SHOWING 1-10 OF 72 REFERENCES

Unsupervised Learning for Real-World Super-Resolution

TLDR
This work learns to invert the effects of bicubic downsampling in order to restore the natural image characteristics present in the data, and can be trained with direct pixel-wise supervision in the high resolution domain, while robustly generalizing to real input.

"Zero-Shot" Super-Resolution Using Deep Internal Learning

TLDR
This paper exploits the internal recurrence of information inside a single image, and train a small image-specific CNN at test time, on examples extracted solely from the input image itself, which is the first unsupervised CNN-based SR method.

Real-World Super-Resolution via Kernel Estimation and Noise Injection

TLDR
This work designs a novel degradation framework for real-world images by estimating various blur kernels as well as real noise distributions and proposes a real- world super-resolution model aiming at better perception.

Blind Super-Resolution Kernel Estimation using an Internal-GAN

TLDR
KernelGAN is introduced, an image-specific Internal-GAN, which trains solely on the LR test image at test time, and learns its internal distribution of patches, and leads to state-of-the-art results in Blind-SR when plugged into existing SR algorithms.

Kernel Modeling Super-Resolution on Real Low-Resolution Images

TLDR
The proposed KMSR consists of two stages: a pool of realistic blur-kernels with a generative adversarial network (GAN) and then a super-resolution network with HR and corresponding LR images constructed with the generated kernels that incorporates blur-kernel modeling in the training.

Unsupervised Image Super-Resolution Using Cycle-in-Cycle Generative Adversarial Networks

TLDR
This work proposes a Cycle-in-Cycle network structure with generative adversarial networks (GAN) as the basic component to tackle the single image super-resolution problem in a more general case that the low-/high-resolution pairs and the down-sampling process are unavailable.

Unsupervised Pixel-Level Domain Adaptation with Generative Adversarial Networks

TLDR
This generative adversarial network (GAN)-based method adapts source-domain images to appear as if drawn from the target domain, and outperforms the state-of-the-art on a number of unsupervised domain adaptation scenarios by large margins.

AIM 2019 Challenge on Real-World Image Super-Resolution: Methods and Results

TLDR
The AIM 2019 challenge on real world super-resolution addresses the real world setting, where paired true high and low-resolution images are unavailable, and aims to advance the state-of-the-art and provide a standard benchmark for this newly emerging task.

EnhanceNet: Single Image Super-Resolution Through Automated Texture Synthesis

TLDR
This work proposes a novel application of automated texture synthesis in combination with a perceptual loss focusing on creating realistic textures rather than optimizing for a pixelaccurate reproduction of ground truth images during training to achieve a significant boost in image quality at high magnification ratios.

Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network

TLDR
SRGAN, a generative adversarial network (GAN) for image super-resolution (SR), is presented, to its knowledge, the first framework capable of inferring photo-realistic natural images for 4x upscaling factors and a perceptual loss function which consists of an adversarial loss and a content loss.
...