NTIRE 2021 Challenge on Video Super-Resolution

Son, Sanghyun; Lee, Suyong; Nah, Seungjun; Timofte, Radu; Lee, Kyoung Mu; Chan, Kelvin C. K.; Zhou, Shangchen; Xu, Xiangyu; Loy, Chen Change; Jiang, Boyuan; Lin, Chuming; Dong, Yuchun; Luo, Donghao; Chu, Wenqing; Ji, Xiaozhong; Yang, Siqian; Tai, Ying; Wang, Chengjie; Li, Jilin; Huang, Feiyue; Chen, Chengpeng; Chu, Xiaojie; Zhang, Jie; Lu, Xin; Chen, Liang‐Yu; Lin, Jiacheng; Du, Guodong; Hao, Jia; Zou, Xueyi; Zhang, Qi; Jiang, Lielin; Li, Xin; Zheng, He; Liu, Fanglong; He, Dongliang; Li, Fu; Dang, Qingqing; Yi, Peng; Wang, Zhongyuan; Jiang, Kui; Jiang, Junjun; Ma, Jiayi; Chen, Yuxiang; Wang, Yutong; Liu, Ting; Sun, Qi-Chao; Liang, Huanwei; Li, Yiming; Li, Zekun; Ruan, Zhubo; Shang, Fanjie; Guo, Chao; Li, Haining; Luo, Ruiyan; Shen, Longjie; Zafirouli, Kassiani; Karageorgos, Konstantinos; Konstantoudakis, Konstantinos; Dimou, Anastasios; Daras, Petros; Song, Xiaowei; Xu, Zhuo; Liu, Hanxi; Guo, Mengxi; Li, Junlin; Liu, Yu; Zhu, Ye; Wang, Qing; Zhao, Shijie; Sun, Xingming; Zhan, Gen; Xie, Tangxin; Jia, Yu; Lu, Yunhua; Zhang, Wenhao; Sun, Mengdi; Michelini, Pablo Navarrete; Zhang, Xueheng; Jiang, Hao; Chen, Zhiyu; Chen, Li; Xiong, Zhiwei; Xiao, Zeyu; Xu, Ruikang; Cheng, Zhen; Fu, Xueyang; Song, Fenglong; Luo, Zhipeng; Yao, Yuehan; Dutta, Saikat; Shah, Nisarg; Das, Sourya Dipta; Zhao, Peng; Shi, Yukai; Liu, Hongying; Shang, Fanhua; Liu, Yuanyuan; Chen, Fei; Yu, Fangxu; Gao, Ruisheng; Bai, Ya; Heo, Jeonghwan; Yue, Shijie; Li, Chenghua; Li, Jinjing; Zheng, Qian; Gang, Ruipeng; Song, Ruixia; Wee, Seungwoo; Jeong, Jechang; Li, Chen; Wen, Geyingjie; Chai, Xinning; Song, Li

doi:10.1109/cvprw53098.2021.00026

Cited by 22 publications

(10 citation statements)

References 46 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We have used REDS STSR dataset [31] for training our models. REDS STSR dataset contains a variety of dynamic scenes.…”

Section: Dataset Descriptionmentioning

confidence: 99%

See 1 more Smart Citation

Efficient Space-time Video Super Resolution using Low-Resolution Flow and Mask Upsampling

Dutta¹,

Shah²,

Mittal³

2021

Preprint

Self Cite

View full text Add to dashboard Cite

This paper explores an efficient solution for Spacetime Super-Resolution, aiming to generate High-resolution Slow-motion videos from Low Resolution and Low Frame rate videos. A simplistic solution is the sequential running of Video Super Resolution and Video Frame interpolation models. However, this type of solutions are memory inefficient, have high inference time, and could not make the proper use of space-time relation property. To this extent, we first interpolate in LR space using quadratic modeling. Input LR frames are super-resolved using a state-of-the-art Video Super-Resolution method. Flowmaps and blending mask which are used to synthesize LR interpolated frame is reused in HR space using bilinear upsampling. This leads to a coarse estimate of HR intermediate frame which often contains artifacts along motion boundaries. We use a refinement network to improve the quality of HR intermediate frame via residual learning. Our model is lightweight and performs better than current state-of-the-art models in REDS STSR Validation set.

show abstract

“…We have used REDS STSR dataset [31] for training our models. REDS STSR dataset contains a variety of dynamic scenes.…”

Section: Dataset Descriptionmentioning

confidence: 99%

“…We have participated in NTIRE 2021 Video Super-Resolution Challenge: Track 2 (Spatio-temporal) [31]. A total of 223 participants registered in this competition out of which 26 teams participated in the validation phase and 14 teams entered the test phase.…”

Section: Ntire 2021 Video Super-resolution Challenge: Trackmentioning

confidence: 99%

Efficient Space-time Video Super Resolution using Low-Resolution Flow and Mask Upsampling

Dutta¹,

Shah²,

Mittal³

2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…This challenge is one of the NTIRE 2021 associated challenges: nonhomogeneous dehazing [4], defocus deblurring using dual-pixel [2], depth guided image relighting [16], image deblurring [39], multi-modal aerial view imagery classification [33], learning the super-resolution space [35], quality enhancement of compressed video (this report), video super-resolution [46], perceptual image quality assessment [18], burst super-resolution [5], and high dynamic range imaging [41].…”

Section: Ntire 2021 Challengementioning

confidence: 99%

NTIRE 2021 Challenge on Quality Enhancement of Compressed Video: Methods and Results

Yang¹,

Timofte²,

Liu³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

This paper reviews the first NTIRE challenge on quality enhancement of compressed video, with a focus on the proposed methods and results. In this challenge, the new Large-scale Diverse Video (LDV) dataset is employed. The challenge has three tracks. Tracks 1 and 2 aim at enhancing the videos compressed by HEVC at a fixed QP, while Track 3 is designed for enhancing the videos compressed by x265 at a fixed bit-rate. Besides, the quality enhancement of Tracks 1 and 3 targets at improving the fidelity (PSNR), and Track 2 targets at enhancing the perceptual quality. The three tracks totally attract 482 registrations. In the test phase, 12 teams, 8 teams and 11 teams submitted the final results of Tracks 1, 2 and 3, respectively. The proposed methods and solutions gauge the state-of-the-art of video quality enhancement. The homepage of the challenge: https: //github.com/RenYang-home/NTIRE21_VEnh

show abstract

“…The NTIRE 2021 HDR Challenge aims at stimulating research for computational HDR imaging, and better understand the state-of-the-art landscape for both single and multiple frame HDR processing. It is part of a wide spectrum of associated challenges with the NTIRE 2021 workshop: non-homogeneous dehazing [3], defocus deblurring using dual-pixel [2], depth guided image relighting [10], image deblurring [25], multi-modal aerial view imagery classification [19], learning the super-resolution space [22], quality enhancement of heavily compressed videos [40], video super-resolution [32], perceptual image quality assessment [12], burst super-resolution [4], high dynamic arXiv:2106.01439v1 [cs.CV] 2 Jun 2021 range [26].…”

Section: Introductionmentioning

confidence: 99%

NTIRE 2021 Challenge on High Dynamic Range Imaging: Dataset, Methods and Results

Pérez-Pellitero

Catley-Chandar

Leonardis

et al. 2021

2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

Self Cite

View full text Add to dashboard Cite

This paper reviews the first challenge on high-dynamic range (HDR) imaging that was part of the New Trends in Image Restoration and Enhancement (NTIRE) workshop, held in conjunction with CVPR 2021. This manuscript focuses on the newly introduced dataset, the proposed methods and their results. The challenge aims at estimating a HDR image from one or multiple respective low-dynamic range (LDR) observations, which might suffer from underor over-exposed regions and different sources of noise. The challenge is composed by two tracks: In Track 1 only a single LDR image is provided as input, whereas in Track 2 three differently-exposed LDR images with inter-frame motion are available. In both tracks, the ultimate goal is to achieve the best objective HDR reconstruction in terms of PSNR with respect to a ground-truth image, evaluated both directly and with a canonical tonemapping operation.

show abstract

NTIRE 2021 Challenge on Video Super-Resolution

Cited by 22 publications

References 46 publications

Efficient Space-time Video Super Resolution using Low-Resolution Flow and Mask Upsampling

Efficient Space-time Video Super Resolution using Low-Resolution Flow and Mask Upsampling

NTIRE 2021 Challenge on Quality Enhancement of Compressed Video: Methods and Results

NTIRE 2021 Challenge on High Dynamic Range Imaging: Dataset, Methods and Results

Contact Info

Product

Resources

About