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

Gu, Shuhang; Danelljan, Martin; Timofte, Radu; Haris, Muhammad; Akita, Kazutoshi; Shakhnarovic, Greg; Ukita, Norimichi; Michelini, Pablo Navarrete; Chen, Wenbin; Liu, Hanwen; Zhu, Dan; Xie, Tangxin; Yang, Xin; Zhu, Chen; Jia, Yu; Sun, Wei; Xin, Tao; Deng, Zijun; Lu, Liying; Li, Wenbo; Guo, Taian; Shen, Xiaoyong; Xu, Xuemiao; Tai, Yu-Wing; Jia, Jiaya; Yi, Peng; Wang, Zhongyuan; Jiang, Kui; Jiang, Junjun; Ma, Jiayi; Liu, Zhisong; Wang, Liwen; Li, Chu-Tak; Siu, Wan-Chi; Chan, Yui-Lam; Zhou, Ruofan; Helou, Majed Ei; Purohit, Kuldeep; Kandula, Praveen; Suin, Maitreya; A.N, Rajagopalan

doi:10.1109/iccvw.2019.00440

Cited by 30 publications

(23 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We increase the number of receptive fields and apply multi-scale self-attention strategy to handle various types of masks in the wild. Our multi-scale self-attention strategy is derived from the non-local network [29], in which the correlation between features is emphasized and it has been used in image super-resolution [6,18]. For achieving better coherency of the completed images, we also adopt the back projection technique [7,18] to encourage better alignment of the generated and valid pixels.…”

Section: Related Workmentioning

confidence: 99%

“…Inspired by the recent success of deep learning approaches at the tasks of image recognition [29,8], image super-resolution [6,19,30], visual place recognition and localization [16,2], image enlightening [27], image synthesis [11,28] and many others, a growing number of CNN based methods of image inpainting [23,32,10,33,17,21,34] have been proposed to fill images with holes in an endto-end manner. For example, Iizuka et al [10] employed dilated convolutions Fig.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

DeepGIN: Deep Generative Inpainting Network for Extreme Image Inpainting

Siu

Liu

et al. 2020

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

DeepGIN: Deep Generative Inpainting Network for Extreme Image Inpainting

Siu

Liu

et al. 2020

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

“…Extreme single image super-resolution reconstruction aims to recover lost high-frequency (rich detail) while maintaining content consistency [13]. Most SR network architectures are designed based on improving the PSNR (Peak Signal-to-Noise Ratio) value.…”

Section: Super Resolution Network With Receptive Field Blockmentioning

confidence: 99%

“…Discriminator loss function of RFB-ESRGAN contains two terms: Real Loss L Real for encouraging the real image is more realistic than fake image, shown as (12). Fake loss L F ake for encouraging the fake image is less realistic than real image, shown as equation (13).…”

Section: Loss Functionmentioning

confidence: 99%

Perceptual Extreme Super Resolution Network with Receptive Field Block

Shang¹,

Dai²,

Zhu³

et al. 2020

Preprint

View full text Add to dashboard Cite

Perceptual Extreme Super-Resolution for single image is extremely difficult, because the texture details of different images vary greatly. To tackle this difficulty, we develop a super resolution network with receptive field block based on Enhanced SRGAN. We call our network RFB-ESRGAN.The key contributions are listed as follows. First, for the purpose of extracting multi-scale information and enhance the feature discriminability, we applied receptive field block (RFB) to super resolution. RFB has achieved competitive results in object detection and classification. Second, instead of using large convolution kernels in multi-scale receptive field block, several small kernels are used in RFB, which makes us be able to extract detailed features and reduce the computation complexity. Third, we alternately use different upsampling methods in the upsampling stage to reduce the high computation complexity and still remain satisfactory performance. Fourth, we use the ensemble of 10 models of different iteration to improve the robustness of model and reduce the noise introduced by each individual model. Our experimental results show the superior performance of RFB-ESRGAN. According to the preliminary results of NTIRE 2020 Perceptual Extreme Super-Resolution Challenge, our solution ranks first among all the participants.

show abstract

“…It can digest extensive training data and extract discriminative representations with the support of powerful computational resource. CNN has shown significant advantages in various tasks, like image classification [18,31], semantic segmentation [28,39], super-resolution [9,24], place recognition [1,20], etc. CNNs with the deep structure are difficult to train because parameters of the shallow layers are often under gradient vanishing and exploding risks.…”

Section: Introductionmentioning

confidence: 99%

Deep Relighting Networks for Image Light Source Manipulation

Wang

Siu

Liu

et al. 2020

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

Manipulating the light source of given images is an interesting task and useful in various applications, including photography and cinematography. Existing methods usually require additional information like the geometric structure of the scene, which may not be available for most images. In this paper, we formulate the single image relighting task and propose a novel Deep Relighting Network (DRN) with three parts: 1) scene reconversion, which aims to reveal the primary scene structure through a deep auto-encoder network, 2) shadow prior estimation, to predict light effect from the new light direction through adversarial learning, and 3) re-renderer, to combine the primary structure with the reconstructed shadow view to form the required estimation under the target light source. Experiments show that the proposed method outperforms other possible methods, both qualitatively and quantitatively. Specifically, the proposed DRN has achieved the best PSNR in the "AIM2020 -Any to one relighting challenge" of the 2020 ECCV conference.

show abstract

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

Cited by 30 publications

References 15 publications

DeepGIN: Deep Generative Inpainting Network for Extreme Image Inpainting

DeepGIN: Deep Generative Inpainting Network for Extreme Image Inpainting

Perceptual Extreme Super Resolution Network with Receptive Field Block

Deep Relighting Networks for Image Light Source Manipulation

Contact Info

Product

Resources

About