{"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/multi-frame-quality-enhancement-for","title":"Multi-Frame Quality Enhancement for Compressed Video","arxiv_id":"1803.04680","date":"2018-03-13","proceeding":"CVPR 2018 6","authors":["Ren Yang","Mai Xu","Zulin Wang","Tianyi Li"],"abstract":"The past few years have witnessed great success in applying deep learning to\nenhance the quality of compressed image/video. The existing approaches mainly\nfocus on enhancing the quality of a single frame, ignoring the similarity\nbetween consecutive frames. In this paper, we investigate that heavy quality\nfluctuation exists across compressed video frames, and thus low quality frames\ncan be enhanced using the neighboring high quality frames, seen as Multi-Frame\nQuality Enhancement (MFQE). Accordingly, this paper proposes an MFQE approach\nfor compressed video, as a first attempt in this direction. In our approach, we\nfirstly develop a Support Vector Machine (SVM) based detector to locate Peak\nQuality Frames (PQFs) in compressed video. Then, a novel Multi-Frame\nConvolutional Neural Network (MF-CNN) is designed to enhance the quality of\ncompressed video, in which the non-PQF and its nearest two PQFs are as the\ninput. The MF-CNN compensates motion between the non-PQF and PQFs through the\nMotion Compensation subnet (MC-subnet). Subsequently, the Quality Enhancement\nsubnet (QE-subnet) reduces compression artifacts of the non-PQF with the help\nof its nearest PQFs. Finally, the experiments validate the effectiveness and\ngenerality of our MFQE approach in advancing the state-of-the-art quality\nenhancement of compressed video. The code of our MFQE approach is available at\nhttps://github.com/ryangBUAA/MFQE.git","url_abs":"http://arxiv.org/abs/1803.04680v4","url_pdf":"http://arxiv.org/pdf/1803.04680v4.pdf","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":"multi-frame-quality-enhancement-for","repo_url":"https://github.com/ryangBUAA/MFQE","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok"}}],"tasks":[{"task_slug":"motion-compensation","task_name":"Motion Compensation"},{"task_slug":"video-enhancement","task_name":"Video Enhancement"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/video-enhancement-on-mfqe-v2","task":"Video Enhancement","dataset":"MFQE v2","model":"MFQE 1.0","rank_in_archive_order":6,"of":6,"metrics":{"Incremental PSNR":"0.46","Parameters(M)":"1.79"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1803.04680","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}