{"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/retinexformer-one-stage-retinex-based","title":"Retinexformer: One-stage Retinex-based Transformer for Low-light Image Enhancement","arxiv_id":"2303.06705","date":"2023-03-12","proceeding":"ICCV 2023 1","authors":["Yuanhao Cai","Hao Bian","Jing Lin","Haoqian Wang","Radu Timofte","Yulun Zhang"],"abstract":"When enhancing low-light images, many deep learning algorithms are based on the Retinex theory. However, the Retinex model does not consider the corruptions hidden in the dark or introduced by the light-up process. Besides, these methods usually require a tedious multi-stage training pipeline and rely on convolutional neural networks, showing limitations in capturing long-range dependencies. In this paper, we formulate a simple yet principled One-stage Retinex-based Framework (ORF). ORF first estimates the illumination information to light up the low-light image and then restores the corruption to produce the enhanced image. We design an Illumination-Guided Transformer (IGT) that utilizes illumination representations to direct the modeling of non-local interactions of regions with different lighting conditions. By plugging IGT into ORF, we obtain our algorithm, Retinexformer. Comprehensive quantitative and qualitative experiments demonstrate that our Retinexformer significantly outperforms state-of-the-art methods on thirteen benchmarks. The user study and application on low-light object detection also reveal the latent practical values of our method. Code, models, and results are available at https://github.com/caiyuanhao1998/Retinexformer","url_abs":"https://arxiv.org/abs/2303.06705v3","url_pdf":"https://arxiv.org/pdf/2303.06705v3.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":"retinexformer-one-stage-retinex-based","repo_url":"https://github.com/caiyuanhao1998/retinexformer","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"retinexformer-one-stage-retinex-based","repo_url":"https://github.com/dawnlh/awesome-low-light-image-enhancement","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"retinexformer-one-stage-retinex-based","repo_url":"https://github.com/DmitryRyumin/ICCV-2023-Papers","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"retinexformer-one-stage-retinex-based","repo_url":"https://github.com/cmhungsteve/Awesome-Transformer-Attention","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"retinexformer-one-stage-retinex-based","repo_url":"https://github.com/lcybuzz/Low-Level-Vision-Paper-Record","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok"}}],"tasks":[{"task_slug":"image-enhancement","task_name":"Image Enhancement"},{"task_slug":"low-light-image-enhancement","task_name":"Low-Light Image Enhancement"},{"task_slug":"low-light-image-deblurring-and-enhancement","task_name":"Low-light Image Deblurring and Enhancement"},{"task_slug":"object-detection","task_name":"Object Detection"},{"task_slug":"photo-retouching","task_name":"Photo Retouching"},{"task_slug":"object-detection-1","task_name":"object-detection"}],"methods":[{"method_slug":"absolute-position-encodings","method_name":"Absolute Position Encodings"},{"method_slug":"adam","method_name":"Adam"},{"method_slug":"attention","method_name":"Attention"},{"method_slug":"bpe","method_name":"BPE"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"label-smoothing","method_name":"Label Smoothing"},{"method_slug":"layer-normalization","method_name":"Layer Normalization"},{"method_slug":"linear-layer","method_name":"Linear Layer"},{"method_slug":"multi-head-attention","method_name":"Multi-Head Attention"},{"method_slug":"position-wise-feed-forward-layer","method_name":"Position-Wise Feed-Forward Layer"},{"method_slug":"residual-connection","method_name":"Residual Connection"},{"method_slug":"softmax","method_name":"Softmax"},{"method_slug":"transformer","method_name":"Transformer"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/image-enhancement-on-mit-adobe-5k","task":"Image Enhancement","dataset":"MIT-Adobe 5k","model":"Retinexformer","rank_in_archive_order":1,"of":11,"metrics":{"PSNR on proRGB":"25.98","PSNR on sRGB":"24.94","SSIM on proRGB":"0.957","SSIM on sRGB":"0.907"},"uses_additional_data":false},{"leaderboard":"/sota/low-light-image-enhancement-on-dicm","task":"Low-Light Image Enhancement","dataset":"DICM","model":"Retinexformer","rank_in_archive_order":5,"of":6,"metrics":{"User Study Score":"3.71"},"uses_additional_data":false},{"leaderboard":"/sota/low-light-image-enhancement-on-lime","task":"Low-Light Image Enhancement","dataset":"LIME","model":"Rextinexformer","rank_in_archive_order":6,"of":6,"metrics":{"User Study Score":"4.3"},"uses_additional_data":false},{"leaderboard":"/sota/low-light-image-enhancement-on-lol","task":"Low-Light Image Enhancement","dataset":"LOL","model":"Retinexformer_","rank_in_archive_order":9,"of":40,"metrics":{"Average PSNR":"27.18","FLOPS (G)":"15.57","Params (M)":"1.61","SSIM":"0.850"},"uses_additional_data":false},{"leaderboard":"/sota/low-light-image-enhancement-on-lol","task":"Low-Light Image Enhancement","dataset":"LOL","model":"Retinexformer","rank_in_archive_order":18,"of":40,"metrics":{"Average PSNR":"25.16","FLOPS (G)":"15.57","Params (M)":"1.61","SSIM":"0.845"},"uses_additional_data":false},{"leaderboard":"/sota/low-light-image-enhancement-on-lol-v2","task":"Low-Light Image Enhancement","dataset":"LOL-v2","model":"Retinexformer","rank_in_archive_order":3,"of":3,"metrics":{"Average PSNR":"22.80","SSIM":"0.840"},"uses_additional_data":false},{"leaderboard":"/sota/low-light-image-enhancement-on-lol-v2-1","task":"Low-Light Image Enhancement","dataset":"LOL-v2-synthetic","model":"Retinexformer","rank_in_archive_order":2,"of":2,"metrics":{"PSNR":"25.67","SSIM":"0.939"},"uses_additional_data":false},{"leaderboard":"/sota/low-light-image-enhancement-on-lolv2","task":"Low-Light Image Enhancement","dataset":"LOLv2","model":"Retinexformer","rank_in_archive_order":10,"of":12,"metrics":{"Average PSNR":"27.71","SSIM":"0.856"},"uses_additional_data":false},{"leaderboard":"/sota/low-light-image-enhancement-on-lolv2-1","task":"Low-Light Image Enhancement","dataset":"LOLv2-synthetic","model":"Retinexformer","rank_in_archive_order":6,"of":9,"metrics":{"Average PSNR":"29.04","SSIM":"0.939"},"uses_additional_data":false},{"leaderboard":"/sota/low-light-image-enhancement-on-mef","task":"Low-Light Image Enhancement","dataset":"MEF","model":"Retinexformer","rank_in_archive_order":6,"of":7,"metrics":{"User Study Score":"3.91"},"uses_additional_data":false},{"leaderboard":"/sota/low-light-image-enhancement-on-mit-adobe-1","task":"Low-Light Image Enhancement","dataset":"MIT-Adobe FiveK","model":"Retinexformer","rank_in_archive_order":1,"of":1,"metrics":{"PSNR":"24.94","SSIM":"0.907"},"uses_additional_data":false},{"leaderboard":"/sota/low-light-image-enhancement-on-npe","task":"Low-Light Image Enhancement","dataset":"NPE","model":"Retinexformer","rank_in_archive_order":6,"of":6,"metrics":{"User Study Score":"4.17"},"uses_additional_data":false},{"leaderboard":"/sota/low-light-image-enhancement-on-sdsd-indoor","task":"Low-Light Image Enhancement","dataset":"SDSD-indoor","model":"Retinexformer","rank_in_archive_order":1,"of":1,"metrics":{"PSNR":"29.77"},"uses_additional_data":false},{"leaderboard":"/sota/low-light-image-enhancement-on-sdsd-outdoor","task":"Low-Light Image Enhancement","dataset":"SDSD-outdoor","model":"Retinexformer","rank_in_archive_order":1,"of":1,"metrics":{"PSNR":"29.84"},"uses_additional_data":false},{"leaderboard":"/sota/low-light-image-enhancement-on-sid","task":"Low-Light Image Enhancement","dataset":"SID","model":"Retinexformer","rank_in_archive_order":1,"of":2,"metrics":{"PSNR":"24.44","SSIM":"0.680"},"uses_additional_data":false},{"leaderboard":"/sota/low-light-image-enhancement-on-smid","task":"Low-Light Image Enhancement","dataset":"SMID","model":"Retinexformer","rank_in_archive_order":1,"of":1,"metrics":{"PSNR":"29.15"},"uses_additional_data":false},{"leaderboard":"/sota/low-light-image-enhancement-on-vv","task":"Low-Light Image Enhancement","dataset":"VV","model":"Rextinexformer","rank_in_archive_order":6,"of":7,"metrics":{"User Study Score":"3.61"},"uses_additional_data":false},{"leaderboard":"/sota/low-light-image-deblurring-and-enhancement-on","task":"Low-light Image Deblurring and Enhancement","dataset":"LOL-Blur","model":"RetinexFormer","rank_in_archive_order":3,"of":3,"metrics":{"Average PSNR":"22.904","LPIPS":"0.236","SSIM":"0.824"},"uses_additional_data":false},{"leaderboard":"/sota/photo-retouching-on-mit-adobe-5k","task":"Photo Retouching","dataset":"MIT-Adobe 5k","model":"Retinexformer","rank_in_archive_order":3,"of":5,"metrics":{"PSNR":"24.94","SSIM":"0.907"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2303.06705","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2303.06705"}},"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/lcybuzz/Low-Level-Vision-Paper-Record","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/cmhungsteve/Awesome-Transformer-Attention","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/DmitryRyumin/ICCV-2023-Papers","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/caiyuanhao1998/retinexformer","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/dawnlh/awesome-low-light-image-enhancement","reach":{"status":"ok"}}],"summary":{"ran":2,"unverified":8},"by_repo_kind":{"official":{"samples":7,"ran":2,"repositories":1},"listed":{"samples":3,"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":"5c89cceea638e59e","entry":"FeedForward","repo":"caiyuanhao1998/retinexformer","repo_kind":"official","path":"basicsr/models/archs/RetinexFormer_arch.py","file_url":"https://github.com/caiyuanhao1998/retinexformer/blob/HEAD/basicsr/models/archs/RetinexFormer_arch.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":"invariant","behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"5c89cceea638e59e"}},{"code_sha256_prefix":"27a881fc4dee86cb","entry":"IGAB","repo":"caiyuanhao1998/retinexformer","repo_kind":"official","path":"basicsr/models/archs/RetinexFormer_arch.py","file_url":"https://github.com/caiyuanhao1998/retinexformer/blob/HEAD/basicsr/models/archs/RetinexFormer_arch.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":"invariant","behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"27a881fc4dee86cb"}},{"code_sha256_prefix":"88a0cffc0364f3e5","entry":"Denoiser","repo":"caiyuanhao1998/retinexformer","repo_kind":"official","path":"basicsr/models/archs/RetinexFormer_arch.py","file_url":"https://github.com/caiyuanhao1998/retinexformer/blob/HEAD/basicsr/models/archs/RetinexFormer_arch.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":"88a0cffc0364f3e5"}},{"code_sha256_prefix":"2dec8ea2cb73b61b","entry":"IG_MSA","repo":"caiyuanhao1998/retinexformer","repo_kind":"official","path":"basicsr/models/archs/RetinexFormer_arch.py","file_url":"https://github.com/caiyuanhao1998/retinexformer/blob/HEAD/basicsr/models/archs/RetinexFormer_arch.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":"2dec8ea2cb73b61b"}},{"code_sha256_prefix":"b88b4cf1d03671a3","entry":"Illumination_Estimator","repo":"caiyuanhao1998/retinexformer","repo_kind":"official","path":"basicsr/models/archs/RetinexFormer_arch.py","file_url":"https://github.com/caiyuanhao1998/retinexformer/blob/HEAD/basicsr/models/archs/RetinexFormer_arch.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":"b88b4cf1d03671a3"}},{"code_sha256_prefix":"aaab867dd2d8ab55","entry":"RetinexFormer","repo":"caiyuanhao1998/retinexformer","repo_kind":"official","path":"basicsr/models/archs/RetinexFormer_arch.py","file_url":"https://github.com/caiyuanhao1998/retinexformer/blob/HEAD/basicsr/models/archs/RetinexFormer_arch.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":"aaab867dd2d8ab55"}},{"code_sha256_prefix":"3aa47abaeb92d33e","entry":"RetinexFormer_Single_Stage","repo":"caiyuanhao1998/retinexformer","repo_kind":"official","path":"basicsr/models/archs/RetinexFormer_arch.py","file_url":"https://github.com/caiyuanhao1998/retinexformer/blob/HEAD/basicsr/models/archs/RetinexFormer_arch.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":"3aa47abaeb92d33e"}},{"code_sha256_prefix":"77db27612c662071","entry":"is_digits","repo":"DmitryRyumin/ICCV-2023-Papers","repo_kind":"listed","path":"code/markdown_to_json_parser.py","file_url":"https://github.com/DmitryRyumin/ICCV-2023-Papers/blob/HEAD/code/markdown_to_json_parser.py","link_basis":"harvester_set","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":"77db27612c662071"}},{"code_sha256_prefix":"0a5cea782a36a0c4","entry":"print_colored_count","repo":"DmitryRyumin/ICCV-2023-Papers","repo_kind":"listed","path":"code/markdown_to_json_parser.py","file_url":"https://github.com/DmitryRyumin/ICCV-2023-Papers/blob/HEAD/code/markdown_to_json_parser.py","link_basis":"harvester_set","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":"0a5cea782a36a0c4"}},{"code_sha256_prefix":"87eceb97e343f3a9","entry":"print_colored_status","repo":"DmitryRyumin/ICCV-2023-Papers","repo_kind":"listed","path":"code/markdown_to_json_parser.py","file_url":"https://github.com/DmitryRyumin/ICCV-2023-Papers/blob/HEAD/code/markdown_to_json_parser.py","link_basis":"harvester_set","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":"87eceb97e343f3a9"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}