{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/real-world-super-resolution-via-kernel","title":"Real-World Super-Resolution via Kernel Estimation and Noise Injection","arxiv_id":null,"date":"2020-06-19","proceeding":"CVPRW 2020 6","authors":["Xiaozhong Ji","Yun Cao","Ying Tai","Chengjie Wang","Jilin Li","Feiyue Huang"],"abstract":"Recent state-of-the-art super-resolution methods have achieved impressive performance on ideal datasets regardless of blur and noise. However, these methods always fail in real-world image super-resolution, since most of them adopt simple bicubic downsampling from high-quality images to construct Low-Resolution (LR) and High-Resolution (HR) pairs for training which may lose track of frequency-related details. To address this issue, we focus on designing a novel degradation framework for real- world images by estimating various blur kernels as well as real noise distributions. Based on our novel degradation framework, we can acquire LR images sharing a common domain with real-world images. Then, we propose a real- world super-resolution model aiming at better perception. Extensive experiments on synthetic noise data and real- world images demonstrate that our method outperforms the state-of-the-art methods, resulting in lower noise and better visual quality. In addition, our method is the winner of NTIRE 2020 Challenge on both tracks of Real-World Super-Resolution, which significantly outperforms other competitors by large margins.","url_abs":"https://ieeexplore.ieee.org/document/9150628","url_pdf":"https://ieeexplore.ieee.org/document/9150628","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"real-world-super-resolution-via-kernel","repo_url":"https://github.com/jixiaozhong/RealSR","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"real-world-super-resolution-via-kernel","repo_url":"https://github.com/nihui/realsr-ncnn-vulkan","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"image-super-resolution","task_name":"Image Super-Resolution"},{"task_slug":"super-resolution","task_name":"Super-Resolution"},{"task_slug":"video-super-resolution","task_name":"Video Super-Resolution"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/video-super-resolution-on-msu-super-1","task":"Video Super-Resolution","dataset":"MSU Super-Resolution for Video Compression","model":"RealSR + x264","rank_in_archive_order":1,"of":85,"metrics":{"BSQ-rate over ERQA":"0.77","BSQ-rate over LPIPS":"0.591","BSQ-rate over MS-SSIM":"0.487","BSQ-rate over PSNR":"0.675","BSQ-rate over Subjective Score":"0.196","BSQ-rate over VMAF":"0.775"},"uses_additional_data":false},{"leaderboard":"/sota/video-super-resolution-on-msu-super-1","task":"Video Super-Resolution","dataset":"MSU Super-Resolution for Video Compression","model":"RealSR + x265","rank_in_archive_order":7,"of":85,"metrics":{"BSQ-rate over ERQA":"1.622","BSQ-rate over LPIPS":"1.206","BSQ-rate over MS-SSIM":"1.033","BSQ-rate over PSNR":"1.064","BSQ-rate over Subjective Score":"0.502","BSQ-rate over VMAF":"1.617"},"uses_additional_data":false},{"leaderboard":"/sota/video-super-resolution-on-msu-super-1","task":"Video Super-Resolution","dataset":"MSU Super-Resolution for Video Compression","model":"RealSR + uavs3e","rank_in_archive_order":8,"of":85,"metrics":{"BSQ-rate over ERQA":"1.943","BSQ-rate over LPIPS":"1.149","BSQ-rate over MS-SSIM":"1.441","BSQ-rate over PSNR":"14.741","BSQ-rate over Subjective Score":"0.639","BSQ-rate over VMAF":"2.253"},"uses_additional_data":false},{"leaderboard":"/sota/video-super-resolution-on-msu-super-1","task":"Video Super-Resolution","dataset":"MSU Super-Resolution for Video Compression","model":"RealSR + aomenc","rank_in_archive_order":15,"of":85,"metrics":{"BSQ-rate over ERQA":"6.762","BSQ-rate over LPIPS":"10.915","BSQ-rate over MS-SSIM":"5.463","BSQ-rate over PSNR":"15.144","BSQ-rate over Subjective Score":"0.843","BSQ-rate over VMAF":"4.283"},"uses_additional_data":false},{"leaderboard":"/sota/video-super-resolution-on-msu-super-1","task":"Video Super-Resolution","dataset":"MSU Super-Resolution for Video Compression","model":"RealSR + vvenc","rank_in_archive_order":85,"of":85,"metrics":{"BSQ-rate over ERQA":"21.965","BSQ-rate over LPIPS":"18.344","BSQ-rate over MS-SSIM":"11.643","BSQ-rate over PSNR":"15.144","BSQ-rate over VMAF":"10.67"},"uses_additional_data":false},{"leaderboard":"/sota/video-super-resolution-on-msu-vsr-benchmark","task":"Video Super-Resolution","dataset":"MSU Video Super Resolution Benchmark: Detail Restoration","model":"RealSR","rank_in_archive_order":17,"of":32,"metrics":{"1 - LPIPS":"0.911","ERQAv1.0":"0.69","FPS":"0.352","PSNR":"25.989","QRCRv1.0":"0","SSIM":"0.767","Subjective score":"5.286"},"uses_additional_data":false},{"leaderboard":"/sota/video-super-resolution-on-msu-video-upscalers","task":"Video Super-Resolution","dataset":"MSU Video Upscalers: Quality Enhancement","model":"RealSR","rank_in_archive_order":15,"of":48,"metrics":{"LPIPS":"0.220","PSNR":"30.64","SSIM":"0.900"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}