AIM 2020 Challenge on Real Image Super-Resolution: Methods and Results

Wei, Pengxu; Lu, Haibao; Timofte, Radu; Li, Lin; Zuo, Wangmeng; Pan, Zhihong; Li, Baopu; Xi, Teng; Fan, Yanwen; Zhang, Gang; Liu, Jingtuo; Han, Junyu; Ding, Errui; Xie, Tangxin; Cao, Liang; Zou, Yan; Shen, Yi; Zhang, Jialiang; Jia, Yu; Cheng, Kaihua; Wu, Chenhuan; Yue, Lin; Liu, Cen; Peng, Yunbo; Zou, Xueyi; Luo, Zhipeng; Yao, Yuehan; Xu, Zhenyu; Zamir, Syed Waqas; Arora, Aditya; Khan, Salman; Hayat, Munawar; Khan, Fahad Shahbaz; Ahn, Keon Hee; Kim, Jun Hyuk; Choi, Jun Ho; Lee, Jong Seok; Zhao, Tongtong; Zhao, Shanshan; Han, Yoseob; Kim, Byung Hoon; Baek, Jae Hyun; Wu, Haoning; Xu, Dejia; Zhou, Bo; Guan, Wei; Li, Xiaobo; Wang, Hongzhi; Li, Hao; Zhong, Haoyu; Shi, Yukai; Yang, Zhijing; Yang, Xuemin; Li, Xin; Jin, Xian-Min; Wu, Yan; Pang, Yi; Liu, Sen; Liu, Zhi Song; Wang, Li Wen; Li, Chu Tak; Cani, Marie Paule; Siu, Wan Chi; Zhou, Yuanbo; Umer, Rao Muhammad; Micheloni, Christian; Xiaofeng, Cong,; Gupta, Rajat; Almasri, Feras; Vandamme, Thomas; Debeir, Olivier

doi:10.1007/978-3-030-67070-2_24

Cited by 34 publications

(33 citation statements)

References 51 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Although so many models have been put forward in blind SR, there is still a long way to go since we have only tackled a small set of real-world images. Existing methods often claim to focus on real-world settings, but they actually assume a certain scene, like images taken by some digital cameras [17], [18]. In fact, real-world images are greatly different in their underlying degradation types, and an SR model designed for a specific type can easily fail for another.…”

Section: Arbitrary Lr Input Domain Gapmentioning

confidence: 99%

“…This kind of approaches aim to implicitly grasp the underlying degradation model through learning with external dataset. For dataset with paired HR-LR images, supervised learning with cautious design of SR network may be already enough to achieve satisfactory results, just like the top solutions proposed in NTIRE 2018 [71] and AIM 2020 [18] challenges. A more difficult setting is learning with unpaired data, where ground truth of LR images with realistic degradations are unavailable.…”

Section: Implicit Degradation Modelling 71 Learning Data Distribution...mentioning

confidence: 99%

See 1 more Smart Citation

Blind Image Super-Resolution: A Survey and Beyond

Liu¹,

Liu²,

Gu³

et al. 2021

Preprint

View full text Add to dashboard Cite

Blind image super-resolution (SR), aiming to super-resolve low-resolution images with unknown degradation, has attracted increasing attention due to its significance in promoting real-world applications. Many novel and effective solutions have been proposed recently, especially with the powerful deep learning techniques. Despite years of efforts, it still remains as a challenging research problem. This paper serves as a systematic review on recent progress in blind image SR, and proposes a taxonomy to categorize existing methods into three different classes according to their ways of degradation modelling and the data used for solving the SR model. This taxonomy helps summarize and distinguish among existing methods. We hope to provide insights into current research states, as well as to reveal novel research directions worth exploring. In addition, we make a summary on commonly used datasets and previous competitions related to blind image SR. Last but not least, a comparison among different methods is provided with detailed analysis on their merits and demerits using both synthetic and real testing images.

show abstract

Section: Arbitrary Lr Input Domain Gapmentioning

confidence: 99%

Section: Implicit Degradation Modelling 71 Learning Data Distribution...mentioning

confidence: 99%

Blind Image Super-Resolution: A Survey and Beyond

Liu¹,

Liu²,

Gu³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…In VCBP, the iterative error correction happens in the low-resolution space, and in each iteration, the reconstructed residual features are added to the HR encoded feature space. In VCBPv2 [21] the parameters are not shared within the modules, and the iterative error correction happens in both lowand high-resolution space. It follows the design of Haris et al [19] of using iterative up and downsampling layers to process the features in the Inner Loop.…”

Section: Related Workmentioning

confidence: 99%

XCycles Backprojection Acoustic Super-Resolution

Almasri

Vandendriessche

Segers

et al. 2021

Sensors

Self Cite

View full text Add to dashboard Cite

The computer vision community has paid much attention to the development of visible image super-resolution (SR) using deep neural networks (DNNs) and has achieved impressive results. The advancement of non-visible light sensors, such as acoustic imaging sensors, has attracted much attention, as they allow people to visualize the intensity of sound waves beyond the visible spectrum. However, because of the limitations imposed on acquiring acoustic data, new methods for improving the resolution of the acoustic images are necessary. At this time, there is no acoustic imaging dataset designed for the SR problem. This work proposed a novel backprojection model architecture for the acoustic image super-resolution problem, together with Acoustic Map Imaging VUB-ULB Dataset (AMIVU). The dataset provides large simulated and real captured images at different resolutions. The proposed XCycles BackProjection model (XCBP), in contrast to the feedforward model approach, fully uses the iterative correction procedure in each cycle to reconstruct the residual error correction for the encoded features in both low- and high-resolution space. The proposed approach was evaluated on the dataset and showed high outperformance compared to the classical interpolation operators and to the recent feedforward state-of-the-art models. It also contributed to a drastically reduced sub-sampling error produced during the data acquisition.

show abstract

“…Despite their success, the previously mentioned methods are trained with LR/HR image pairs with the bicubic downsampling and thus they have limited performance in the real-world settings. Recently, in the real-world SR challenge series [18]- [20], the authors have described the effects of bicubic downsampling. More recently, in [1], [2], the authors proposed GAN-based SR methods to solve the real-world SR problem.…”

Section: Related Workmentioning

confidence: 99%

“…For the training, we use DIV2K [22], Flickr2K [23], and RealSR [24] datasets that jointly contain 22,430 high-quality HR images for image restoration tasks with rich and diverse textures. We obtain the LR bicubic, LR bilinear, and LR nearest images by down-sampling HR images by the scaling factor ×4 of the DIV2K and Flickr2K datasets using the Pytorch bicubic, bilinear, and nearest kernel function.…”

Section: A Training Datamentioning

confidence: 99%

A Deep Residual Star Generative Adversarial Network for multi-domain Image Super-Resolution

Umer¹,

Munir²,

Micheloni³

2021

Preprint

Self Cite

View full text Add to dashboard Cite

Recently, most of state-of-the-art single image superresolution (SISR) methods have attained impressive performance by using deep convolutional neural networks (DCNNs). The existing SR methods have limited performance due to a fixed degradation settings, i.e. usually a bicubic downscaling of lowresolution (LR) image. However, in real-world settings, the LR degradation process is unknown which can be bicubic LR, bilinear LR, nearest-neighbor LR, or real LR. Therefore, most SR methods are ineffective and inefficient in handling more than one degradation settings within a single network. To handle the multiple degradation, i.e. refers to multi-domain image super-resolution, we propose a deep Super-Resolution Residual StarGAN (SR 2 * GAN), a novel and scalable approach that superresolves the LR images for the multiple LR domains using only a single model. The proposed scheme is trained in a StarGAN like network topology with a single generator and discriminator networks. We demonstrate the effectiveness of our proposed approach in quantitative and qualitative experiments compared to other state-of-the-art methods.

show abstract

AIM 2020 Challenge on Real Image Super-Resolution: Methods and Results

Cited by 34 publications

References 51 publications

Blind Image Super-Resolution: A Survey and Beyond

Blind Image Super-Resolution: A Survey and Beyond

XCycles Backprojection Acoustic Super-Resolution

A Deep Residual Star Generative Adversarial Network for multi-domain Image Super-Resolution

Contact Info

Product

Resources

About