NTIRE 2022 Challenge on Efficient Super-Resolution: Methods and Results
Abstract
This paper reviews the NTIRE 2022 challenge on efficient single image super-resolution with focus on the proposed solutions and results. The task of the challenge was to super-resolve an input image with a magnification factor of 4 based on pairs of low and corresponding high resolution images. The aim was to design a network for single image super-resolution that achieved improvement of efficiency measured according to several metrics including runtime, parameters, FLOPs, activations, and memory consumption while at least maintaining the PSNR of 29.00dB on DIV2K validation set. IMDN is set as the baseline for efficiency measurement. The challenge had 3 tracks including the main track (runtime), sub-track one (model complexity), and sub-track two (overall performance). In the main track, the practical runtime performance of the submissions was evaluated. The rank of the teams were determined directly by the absolute value of the average runtime on the validation set and test set. In sub-track one, the number of parameters and FLOPs were considered. And the individual rankings of the two metrics were summed up to determine a final ranking in this track. In sub-track two, all of the five metrics mentioned in the description of the challenge including runtime, parameter count, FLOPs, activations, and memory consumption were considered. Similar to sub-track one, the rankings of five metrics were summed up to determine a final ranking. The challenge had 303 registered participants, and 43 teams made valid submissions. They gauge the state-of-the-art in efficient single image super-resolution.
Keywords
Cite
@article{arxiv.2205.05675,
title = {NTIRE 2022 Challenge on Efficient Super-Resolution: Methods and Results},
author = {Yawei Li and Kai Zhang and Radu Timofte and Luc Van Gool and Fangyuan Kong and Mingxi Li and Songwei Liu and Zongcai Du and Ding Liu and Chenhui Zhou and Jingyi Chen and Qingrui Han and Zheyuan Li and Yingqi Liu and Xiangyu Chen and Haoming Cai and Yu Qiao and Chao Dong and Long Sun and Jinshan Pan and Yi Zhu and Zhikai Zong and Xiaoxiao Liu and Zheng Hui and Tao Yang and Peiran Ren and Xuansong Xie and Xian-Sheng Hua and Yanbo Wang and Xiaozhong Ji and Chuming Lin and Donghao Luo and Ying Tai and Chengjie Wang and Zhizhong Zhang and Yuan Xie and Shen Cheng and Ziwei Luo and Lei Yu and Zhihong Wen and Qi Wu1 and Youwei Li and Haoqiang Fan and Jian Sun and Shuaicheng Liu and Yuanfei Huang and Meiguang Jin and Hua Huang and Jing Liu and Xinjian Zhang and Yan Wang and Lingshun Long and Gen Li and Yuanfan Zhang and Zuowei Cao and Lei Sun and Panaetov Alexander and Yucong Wang and Minjie Cai and Li Wang and Lu Tian and Zheyuan Wang and Hongbing Ma and Jie Liu and Chao Chen and Yidong Cai and Jie Tang and Gangshan Wu and Weiran Wang and Shirui Huang and Honglei Lu and Huan Liu and Keyan Wang and Jun Chen and Shi Chen and Yuchun Miao and Zimo Huang and Lefei Zhang and Mustafa Ayazoğlu and Wei Xiong and Chengyi Xiong and Fei Wang and Hao Li and Ruimian Wen and Zhijing Yang and Wenbin Zou and Weixin Zheng and Tian Ye and Yuncheng Zhang and Xiangzhen Kong and Aditya Arora and Syed Waqas Zamir and Salman Khan and Munawar Hayat and Fahad Shahbaz Khan and Dandan Gaoand Dengwen Zhouand Qian Ning and Jingzhu Tang and Han Huang and Yufei Wang and Zhangheng Peng and Haobo Li and Wenxue Guan and Shenghua Gong and Xin Li and Jun Liu and Wanjun Wang and Dengwen Zhou and Kun Zeng and Hanjiang Lin and Xinyu Chen and Jinsheng Fang},
journal= {arXiv preprint arXiv:2205.05675},
year = {2022}
}
Comments
Validation code of the baseline model is available at https://github.com/ofsoundof/IMDN. Validation of all submitted models is available at https://github.com/ofsoundof/NTIRE2022_ESR