{"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/frame-recurrent-video-super-resolution","title":"Frame-Recurrent Video Super-Resolution","arxiv_id":"1801.04590","date":"2018-01-14","proceeding":"CVPR 2018 6","authors":["Mehdi S. M. Sajjadi","Raviteja Vemulapalli","Matthew Brown"],"abstract":"Recent advances in video super-resolution have shown that convolutional\nneural networks combined with motion compensation are able to merge information\nfrom multiple low-resolution (LR) frames to generate high-quality images.\nCurrent state-of-the-art methods process a batch of LR frames to generate a\nsingle high-resolution (HR) frame and run this scheme in a sliding window\nfashion over the entire video, effectively treating the problem as a large\nnumber of separate multi-frame super-resolution tasks. This approach has two\nmain weaknesses: 1) Each input frame is processed and warped multiple times,\nincreasing the computational cost, and 2) each output frame is estimated\nindependently conditioned on the input frames, limiting the system's ability to\nproduce temporally consistent results.\n  In this work, we propose an end-to-end trainable frame-recurrent video\nsuper-resolution framework that uses the previously inferred HR estimate to\nsuper-resolve the subsequent frame. This naturally encourages temporally\nconsistent results and reduces the computational cost by warping only one image\nin each step. Furthermore, due to its recurrent nature, the proposed method has\nthe ability to assimilate a large number of previous frames without increased\ncomputational demands. Extensive evaluations and comparisons with previous\nmethods validate the strengths of our approach and demonstrate that the\nproposed framework is able to significantly outperform the current state of the\nart.","url_abs":"http://arxiv.org/abs/1801.04590v4","url_pdf":"http://arxiv.org/pdf/1801.04590v4.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":[],"tasks":[{"task_slug":"motion-compensation","task_name":"Motion Compensation"},{"task_slug":"multi-frame-super-resolution","task_name":"Multi-Frame 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-video-upscalers","task":"Video Super-Resolution","dataset":"MSU Video Upscalers: Quality Enhancement","model":"FRVSR","rank_in_archive_order":44,"of":48,"metrics":{"PSNR":"27.23","SSIM":"0.936","VMAF":"57.14"},"uses_additional_data":false},{"leaderboard":"/sota/video-super-resolution-on-vid4-4x-upscaling-1","task":"Video Super-Resolution","dataset":"Vid4 - 4x upscaling - BD degradation","model":"FRVSR","rank_in_archive_order":17,"of":18,"metrics":{"PSNR":"26.69","SSIM":"0.8103"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1801.04590","atlas_url":"https://app.syntology.ai/?focus=1801.04590","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}