{"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/non-local-recurrent-network-for-image","title":"Non-Local Recurrent Network for Image Restoration","arxiv_id":"1806.02919","date":"2018-06-07","proceeding":"NeurIPS 2018 12","authors":["Ding Liu","Bihan Wen","Yuchen Fan","Chen Change Loy","Thomas S. Huang"],"abstract":"Many classic methods have shown non-local self-similarity in natural images\nto be an effective prior for image restoration. However, it remains unclear and\nchallenging to make use of this intrinsic property via deep networks. In this\npaper, we propose a non-local recurrent network (NLRN) as the first attempt to\nincorporate non-local operations into a recurrent neural network (RNN) for\nimage restoration. The main contributions of this work are: (1) Unlike existing\nmethods that measure self-similarity in an isolated manner, the proposed\nnon-local module can be flexibly integrated into existing deep networks for\nend-to-end training to capture deep feature correlation between each location\nand its neighborhood. (2) We fully employ the RNN structure for its parameter\nefficiency and allow deep feature correlation to be propagated along adjacent\nrecurrent states. This new design boosts robustness against inaccurate\ncorrelation estimation due to severely degraded images. (3) We show that it is\nessential to maintain a confined neighborhood for computing deep feature\ncorrelation given degraded images. This is in contrast to existing practice\nthat deploys the whole image. Extensive experiments on both image denoising and\nsuper-resolution tasks are conducted. Thanks to the recurrent non-local\noperations and correlation propagation, the proposed NLRN achieves superior\nresults to state-of-the-art methods with much fewer parameters.","url_abs":"http://arxiv.org/abs/1806.02919v2","url_pdf":"http://arxiv.org/pdf/1806.02919v2.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":"non-local-recurrent-network-for-image","repo_url":"https://github.com/Ding-Liu/NLRN","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"denoising","task_name":"Denoising"},{"task_slug":"feature-correlation","task_name":"Feature Correlation"},{"task_slug":"image-denoising","task_name":"Image Denoising"},{"task_slug":"image-restoration","task_name":"Image Restoration"},{"task_slug":"image-super-resolution","task_name":"Image Super-Resolution"},{"task_slug":"super-resolution","task_name":"Super-Resolution"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/denoising-on-darmstadt-noise-dataset","task":"Denoising","dataset":"Darmstadt Noise Dataset","model":"NLRN","rank_in_archive_order":10,"of":10,"metrics":{"PSNR":"30.8"},"uses_additional_data":false},{"leaderboard":"/sota/grayscale-image-denoising-on-bsd200-sigma30","task":"Grayscale Image Denoising","dataset":"BSD200 sigma30","model":"NLRN-MV","rank_in_archive_order":2,"of":3,"metrics":{"PSNR":"28.2"},"uses_additional_data":false},{"leaderboard":"/sota/grayscale-image-denoising-on-bsd200-sigma50","task":"Grayscale Image Denoising","dataset":"BSD200 sigma50","model":"NLRN-MV","rank_in_archive_order":2,"of":3,"metrics":{"PSNR":"25.97"},"uses_additional_data":false},{"leaderboard":"/sota/grayscale-image-denoising-on-bsd200-sigma70","task":"Grayscale Image Denoising","dataset":"BSD200 sigma70","model":"NLRN-MV","rank_in_archive_order":2,"of":3,"metrics":{"PSNR":"24.62"},"uses_additional_data":false},{"leaderboard":"/sota/grayscale-image-denoising-on-bsd68-sigma15","task":"Grayscale Image Denoising","dataset":"BSD68 sigma15","model":"NLRN","rank_in_archive_order":5,"of":16,"metrics":{"PSNR":"31.88"},"uses_additional_data":false},{"leaderboard":"/sota/grayscale-image-denoising-on-bsd68-sigma25","task":"Grayscale Image Denoising","dataset":"BSD68 sigma25","model":"NLRN","rank_in_archive_order":3,"of":16,"metrics":{"PSNR":"29.41"},"uses_additional_data":false},{"leaderboard":"/sota/grayscale-image-denoising-on-bsd68-sigma50","task":"Grayscale Image Denoising","dataset":"BSD68 sigma50","model":"NLRN","rank_in_archive_order":5,"of":15,"metrics":{"PSNR":"26.47"},"uses_additional_data":false},{"leaderboard":"/sota/grayscale-image-denoising-on-set12-sigma15","task":"Grayscale Image Denoising","dataset":"Set12 sigma15","model":"NLRN","rank_in_archive_order":3,"of":8,"metrics":{"PSNR":"33.16"},"uses_additional_data":false},{"leaderboard":"/sota/grayscale-image-denoising-on-set12-sigma30","task":"Grayscale Image Denoising","dataset":"Set12 sigma30","model":"NLRN","rank_in_archive_order":1,"of":2,"metrics":{"PSNR":"30.8"},"uses_additional_data":false},{"leaderboard":"/sota/grayscale-image-denoising-on-set12-sigma50","task":"Grayscale Image Denoising","dataset":"Set12 sigma50","model":"NLRN","rank_in_archive_order":4,"of":8,"metrics":{"PSNR":"27.64"},"uses_additional_data":false},{"leaderboard":"/sota/grayscale-image-denoising-on-urban100-sigma15","task":"Grayscale Image Denoising","dataset":"Urban100 sigma15","model":"NLRN","rank_in_archive_order":4,"of":7,"metrics":{"PSNR":"33.45"},"uses_additional_data":false},{"leaderboard":"/sota/grayscale-image-denoising-on-urban100-sigma25","task":"Grayscale Image Denoising","dataset":"Urban100 sigma25","model":"NLRN","rank_in_archive_order":7,"of":10,"metrics":{"PSNR":"30.94"},"uses_additional_data":false},{"leaderboard":"/sota/grayscale-image-denoising-on-urban100-sigma50","task":"Grayscale Image Denoising","dataset":"Urban100 sigma50","model":"NLRN","rank_in_archive_order":6,"of":10,"metrics":{"PSNR":"27.49"},"uses_additional_data":false},{"leaderboard":"/sota/image-super-resolution-on-bsd100-4x-upscaling","task":"Image Super-Resolution","dataset":"BSD100 - 4x upscaling","model":"NLRN","rank_in_archive_order":38,"of":71,"metrics":{"PSNR":"27.48","SSIM":"0.7306"},"uses_additional_data":false},{"leaderboard":"/sota/image-super-resolution-on-set14-4x-upscaling","task":"Image Super-Resolution","dataset":"Set14 - 4x upscaling","model":"NLRN","rank_in_archive_order":70,"of":104,"metrics":{"PSNR":"28.36","SSIM":"0.7745"},"uses_additional_data":false},{"leaderboard":"/sota/image-super-resolution-on-urban100-4x","task":"Image Super-Resolution","dataset":"Urban100 - 4x upscaling","model":"NLRN","rank_in_archive_order":47,"of":65,"metrics":{"PSNR":"25.79","SSIM":"0.7729"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1806.02919","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"1806.02919"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/Ding-Liu/NLRN","reach":{"status":"ok","spdx":"MIT"}}],"summary":{"unverified":1},"by_repo_kind":{"official":{"samples":1,"ran":0,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":0,"samples":[{"code_sha256_prefix":"a67673ff07558f6c","entry":"model_fn","repo":"Ding-Liu/NLRN","repo_kind":"official","path":"models/nlrn.py","file_url":"https://github.com/Ding-Liu/NLRN/blob/HEAD/models/nlrn.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"a67673ff07558f6c"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}