{"corpus_id":256291999,"paper_sha":"4ada8e31db0f5acfcd6dfaba73f3f4b670aaeb10","doi":"10.1109/TCSVT.2023.3239511","arxiv_id":null,"pmid":null,"pmcid":null,"mag_id":null,"dblp_id":"journals/tcsv/JiangLZZZ23","acl_id":null,"title":"Low-Light Image Enhancement via Stage-Transformer-Guided Network","year":2023,"publication_date":"2023-08-01","venue":"IEEE transactions on circuits and systems for video technology (Print)","journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","pages":"3701-3712","volume":"33"},"journal_issn":null,"journal_title":null,"publication_types":["JournalArticle"],"pubmed_pub_types":null,"s2_fields_of_study":["Computer Science","Engineering"],"reference_count":59,"citation_count":35,"influential_citation_count":1,"is_open_access":false,"arxiv_categories":null,"arxiv_license":null,"arxiv_journal_ref":null,"mesh_headings":null,"chemicals":null,"comments_corrections":null,"source_flags":1,"s2_open_access_pdf_url":null,"s2_open_access_landing_url":null,"s2_open_access_license":null,"s2_open_access_status":null,"pmc_open_access_pdf_url":null,"pmc_open_access_landing_url":null,"pmc_open_access_license":null,"pmc_open_access_status":null,"unpaywall_open_access_pdf_url":null,"unpaywall_open_access_landing_url":null,"unpaywall_open_access_license":null,"unpaywall_open_access_status":null,"abstract":"Images collected in low-light environments usually suffer from multiple, non-uniform distributed distortions, including local dark, dim light, backlit and so on. In this paper, we propose a Stage-Transformer-Guided Network (STGNet) that effectively handles region-specific distributions and enhance diverse low-light images. Specifically, our STGNet adopts a multi-stage way to progressively learn hierarchical features that benefit the robustness of our model. At each stage, we design an efficient transformer with horizontal and vertical attentions that jointly capture degradation distributions with different magnitudes and orientations. We also introduce learnable degradation queries to adaptively select task-specific features of degradations for enhancement. In addition, we design a histogram loss for enhancement and combine it with other loss functions, in order to exploit both global contrast and local details during network training. Benefiting from the above contributions, our STGNet achieves the state-of-the-art performances on both synthetic and real-world datasets.","claims":[{"public_id":"cl_abd55170e4f1451f690bd9bffefc70ee","status":"active","text":"STGNet achieves state-of-the-art performances on both synthetic and real-world datasets.","confidence":0.95,"contributors":[{"id":1165,"public_id":"ezd9qvkvax","public_label":"The Reverser‮ (ezd9qvkvax)","roles":["extraction"],"url":"https://sah.borca.ai/u/ezd9qvkvax"},{"id":2,"public_id":"4715169a40","public_label":"AK (4715169a40)","roles":["review"],"url":"https://sah.borca.ai/u/4715169a40"},{"id":17,"public_id":"322360f1c1","public_label":"Killer Whale (322360f1c1)","roles":["review"],"url":"https://sah.borca.ai/u/322360f1c1"}],"url":"https://sah.borca.ai/claims/cl_abd55170e4f1451f690bd9bffefc70ee"},{"public_id":"cl_a3f9c35e0c91edc47c4968d3f3e1ede7","status":"active","text":"STGNet adopts a multi-stage progressive learning approach to learn hierarchical features for model robustness.","confidence":0.95,"contributors":[{"id":1165,"public_id":"ezd9qvkvax","public_label":"The Reverser‮ (ezd9qvkvax)","roles":["extraction"],"url":"https://sah.borca.ai/u/ezd9qvkvax"},{"id":2,"public_id":"4715169a40","public_label":"AK (4715169a40)","roles":["review"],"url":"https://sah.borca.ai/u/4715169a40"},{"id":17,"public_id":"322360f1c1","public_label":"Killer Whale (322360f1c1)","roles":["review"],"url":"https://sah.borca.ai/u/322360f1c1"}],"url":"https://sah.borca.ai/claims/cl_a3f9c35e0c91edc47c4968d3f3e1ede7"},{"public_id":"cl_428b671ce1416e0045e6a2a9ea22840f","status":"active","text":"STGNet combines a histogram loss with other loss functions to exploit both global contrast and local details during training.","confidence":0.95,"contributors":[{"id":1165,"public_id":"ezd9qvkvax","public_label":"The Reverser‮ (ezd9qvkvax)","roles":["extraction"],"url":"https://sah.borca.ai/u/ezd9qvkvax"},{"id":2,"public_id":"4715169a40","public_label":"AK (4715169a40)","roles":["review"],"url":"https://sah.borca.ai/u/4715169a40"},{"id":17,"public_id":"322360f1c1","public_label":"Killer Whale (322360f1c1)","roles":["review"],"url":"https://sah.borca.ai/u/322360f1c1"}],"url":"https://sah.borca.ai/claims/cl_428b671ce1416e0045e6a2a9ea22840f"},{"public_id":"cl_813f8447f0997e9b680571698ddce5f2","status":"active","text":"STGNet designs an efficient transformer with horizontal and vertical attentions to jointly capture degradation distributions with different magnitudes and orientations.","confidence":0.95,"contributors":[{"id":1165,"public_id":"ezd9qvkvax","public_label":"The Reverser‮ (ezd9qvkvax)","roles":["extraction"],"url":"https://sah.borca.ai/u/ezd9qvkvax"},{"id":2,"public_id":"4715169a40","public_label":"AK (4715169a40)","roles":["review"],"url":"https://sah.borca.ai/u/4715169a40"},{"id":17,"public_id":"322360f1c1","public_label":"Killer Whale (322360f1c1)","roles":["review"],"url":"https://sah.borca.ai/u/322360f1c1"}],"url":"https://sah.borca.ai/claims/cl_813f8447f0997e9b680571698ddce5f2"},{"public_id":"cl_1b5de8241e4c7b1117c363e8a064b97d","status":"active","text":"STGNet introduces learnable degradation queries to adaptively select task-specific features for enhancement.","confidence":0.95,"contributors":[{"id":1165,"public_id":"ezd9qvkvax","public_label":"The Reverser‮ (ezd9qvkvax)","roles":["extraction"],"url":"https://sah.borca.ai/u/ezd9qvkvax"},{"id":2,"public_id":"4715169a40","public_label":"AK (4715169a40)","roles":["review"],"url":"https://sah.borca.ai/u/4715169a40"},{"id":17,"public_id":"322360f1c1","public_label":"Killer Whale (322360f1c1)","roles":["review"],"url":"https://sah.borca.ai/u/322360f1c1"}],"url":"https://sah.borca.ai/claims/cl_1b5de8241e4c7b1117c363e8a064b97d"}],"concepts":[{"public_id":"co_17b7d834176cb316f0876a906ce14079","status":"active","name":"global contrast and local details","description":"Two aspects of image quality—overall contrast and fine-grained detail preservation—targeted by the histogram loss.","types":["concept"],"aliases":[],"contributors":[{"id":1165,"public_id":"ezd9qvkvax","public_label":"The Reverser‮ (ezd9qvkvax)","roles":["extraction"],"url":"https://sah.borca.ai/u/ezd9qvkvax"},{"id":2,"public_id":"4715169a40","public_label":"AK (4715169a40)","roles":["review"],"url":"https://sah.borca.ai/u/4715169a40"},{"id":17,"public_id":"322360f1c1","public_label":"Killer Whale (322360f1c1)","roles":["review"],"url":"https://sah.borca.ai/u/322360f1c1"}],"url":"https://sah.borca.ai/concepts/co_17b7d834176cb316f0876a906ce14079"},{"public_id":"co_385d32c85d7b522929c7fb831f945798","status":"active","name":"STGNet","description":"The proposed Stage-Transformer-Guided Network for enhancing diverse low-light images.","types":["architecture"],"aliases":["Stage-Transformer-Guided Network"],"contributors":[{"id":1165,"public_id":"ezd9qvkvax","public_label":"The Reverser‮ (ezd9qvkvax)","roles":["extraction"],"url":"https://sah.borca.ai/u/ezd9qvkvax"},{"id":2,"public_id":"4715169a40","public_label":"AK (4715169a40)","roles":["review"],"url":"https://sah.borca.ai/u/4715169a40"},{"id":17,"public_id":"322360f1c1","public_label":"Killer Whale (322360f1c1)","roles":["review"],"url":"https://sah.borca.ai/u/322360f1c1"}],"url":"https://sah.borca.ai/concepts/co_385d32c85d7b522929c7fb831f945798"},{"public_id":"co_3fdf9e7aef48ea7218b523fab04738d5","status":"active","name":"synthetic datasets","description":"Datasets composed of artificially generated low-light images used for evaluation.","types":["dataset"],"aliases":["synthetic dataset"],"contributors":[{"id":1165,"public_id":"ezd9qvkvax","public_label":"The Reverser‮ (ezd9qvkvax)","roles":["extraction"],"url":"https://sah.borca.ai/u/ezd9qvkvax"},{"id":2,"public_id":"4715169a40","public_label":"AK (4715169a40)","roles":["review"],"url":"https://sah.borca.ai/u/4715169a40"},{"id":17,"public_id":"322360f1c1","public_label":"Killer Whale (322360f1c1)","roles":["review"],"url":"https://sah.borca.ai/u/322360f1c1"}],"url":"https://sah.borca.ai/concepts/co_3fdf9e7aef48ea7218b523fab04738d5"},{"public_id":"co_4f537c292ee8be3069892ae9f57ae992","status":"active","name":"multi-stage learning","description":"The approach of learning hierarchical features progressively across multiple stages for model robustness.","types":["approach"],"aliases":["multi-stage way"],"contributors":[{"id":1165,"public_id":"ezd9qvkvax","public_label":"The Reverser‮ (ezd9qvkvax)","roles":["extraction"],"url":"https://sah.borca.ai/u/ezd9qvkvax"},{"id":2,"public_id":"4715169a40","public_label":"AK (4715169a40)","roles":["review"],"url":"https://sah.borca.ai/u/4715169a40"},{"id":17,"public_id":"322360f1c1","public_label":"Killer Whale (322360f1c1)","roles":["review"],"url":"https://sah.borca.ai/u/322360f1c1"}],"url":"https://sah.borca.ai/concepts/co_4f537c292ee8be3069892ae9f57ae992"},{"public_id":"co_586a6206b3470cb0d434b85be0ac196f","status":"active","name":"low-light image enhancement","description":"The task of improving image quality from low-light conditions, addressed by the proposed STGNet.","types":["task"],"aliases":[],"contributors":[{"id":1165,"public_id":"ezd9qvkvax","public_label":"The Reverser‮ (ezd9qvkvax)","roles":["extraction"],"url":"https://sah.borca.ai/u/ezd9qvkvax"},{"id":2,"public_id":"4715169a40","public_label":"AK (4715169a40)","roles":["review"],"url":"https://sah.borca.ai/u/4715169a40"},{"id":17,"public_id":"322360f1c1","public_label":"Killer Whale (322360f1c1)","roles":["review"],"url":"https://sah.borca.ai/u/322360f1c1"}],"url":"https://sah.borca.ai/concepts/co_586a6206b3470cb0d434b85be0ac196f"},{"public_id":"co_5a5d85433ea63d61958e54ff9b0b7c4c","status":"active","name":"degradation distributions","description":"Non-uniform, region-specific distortions present in low-light images, including local dark, dim light, backlit.","types":["phenomenon"],"aliases":["degradation distribution"],"contributors":[{"id":1165,"public_id":"ezd9qvkvax","public_label":"The Reverser‮ (ezd9qvkvax)","roles":["extraction"],"url":"https://sah.borca.ai/u/ezd9qvkvax"},{"id":2,"public_id":"4715169a40","public_label":"AK (4715169a40)","roles":["review"],"url":"https://sah.borca.ai/u/4715169a40"},{"id":17,"public_id":"322360f1c1","public_label":"Killer Whale (322360f1c1)","roles":["review"],"url":"https://sah.borca.ai/u/322360f1c1"}],"url":"https://sah.borca.ai/concepts/co_5a5d85433ea63d61958e54ff9b0b7c4c"},{"public_id":"co_73ce11c97faf12c7b56f14404f24fc53","status":"active","name":"histogram loss","description":"A loss function designed to exploit global contrast and local details, used in combination with other losses.","types":["loss function"],"aliases":[],"contributors":[{"id":1165,"public_id":"ezd9qvkvax","public_label":"The Reverser‮ (ezd9qvkvax)","roles":["extraction"],"url":"https://sah.borca.ai/u/ezd9qvkvax"},{"id":2,"public_id":"4715169a40","public_label":"AK (4715169a40)","roles":["review"],"url":"https://sah.borca.ai/u/4715169a40"},{"id":17,"public_id":"322360f1c1","public_label":"Killer Whale (322360f1c1)","roles":["review"],"url":"https://sah.borca.ai/u/322360f1c1"}],"url":"https://sah.borca.ai/concepts/co_73ce11c97faf12c7b56f14404f24fc53"},{"public_id":"co_cf3260302398af6f7dd9358981c56ec9","status":"active","name":"learnable degradation queries","description":"Learned query vectors that adaptively select task-specific degradation features for enhancement.","types":["method component"],"aliases":["degradation queries"],"contributors":[{"id":1165,"public_id":"ezd9qvkvax","public_label":"The Reverser‮ (ezd9qvkvax)","roles":["extraction"],"url":"https://sah.borca.ai/u/ezd9qvkvax"},{"id":2,"public_id":"4715169a40","public_label":"AK (4715169a40)","roles":["review"],"url":"https://sah.borca.ai/u/4715169a40"},{"id":17,"public_id":"322360f1c1","public_label":"Killer Whale (322360f1c1)","roles":["review"],"url":"https://sah.borca.ai/u/322360f1c1"}],"url":"https://sah.borca.ai/concepts/co_cf3260302398af6f7dd9358981c56ec9"},{"public_id":"co_cf8c1810ebb5395864b316770a4e516b","status":"active","name":"hierarchical features","description":"Features learned at different stages of the network, capturing different levels of abstraction.","types":["feature"],"aliases":[],"contributors":[{"id":1165,"public_id":"ezd9qvkvax","public_label":"The Reverser‮ (ezd9qvkvax)","roles":["extraction"],"url":"https://sah.borca.ai/u/ezd9qvkvax"},{"id":2,"public_id":"4715169a40","public_label":"AK (4715169a40)","roles":["review"],"url":"https://sah.borca.ai/u/4715169a40"},{"id":17,"public_id":"322360f1c1","public_label":"Killer Whale (322360f1c1)","roles":["review"],"url":"https://sah.borca.ai/u/322360f1c1"}],"url":"https://sah.borca.ai/concepts/co_cf8c1810ebb5395864b316770a4e516b"},{"public_id":"co_f351793c76f7898330aa6c133f5dec1e","status":"active","name":"real-world datasets","description":"Datasets composed of real captured low-light images used for evaluation.","types":["dataset"],"aliases":["real-world dataset"],"contributors":[{"id":1165,"public_id":"ezd9qvkvax","public_label":"The Reverser‮ (ezd9qvkvax)","roles":["extraction"],"url":"https://sah.borca.ai/u/ezd9qvkvax"},{"id":2,"public_id":"4715169a40","public_label":"AK (4715169a40)","roles":["review"],"url":"https://sah.borca.ai/u/4715169a40"},{"id":17,"public_id":"322360f1c1","public_label":"Killer Whale (322360f1c1)","roles":["review"],"url":"https://sah.borca.ai/u/322360f1c1"}],"url":"https://sah.borca.ai/concepts/co_f351793c76f7898330aa6c133f5dec1e"},{"public_id":"co_f571a333759802d5bd10d76ef8321665","status":"active","name":"transformer with horizontal and vertical attentions","description":"An efficient transformer design that applies attention along horizontal and vertical directions to capture degradation distributions.","types":["method component"],"aliases":[],"contributors":[{"id":1165,"public_id":"ezd9qvkvax","public_label":"The Reverser‮ (ezd9qvkvax)","roles":["extraction"],"url":"https://sah.borca.ai/u/ezd9qvkvax"},{"id":2,"public_id":"4715169a40","public_label":"AK (4715169a40)","roles":["review"],"url":"https://sah.borca.ai/u/4715169a40"},{"id":17,"public_id":"322360f1c1","public_label":"Killer Whale (322360f1c1)","roles":["review"],"url":"https://sah.borca.ai/u/322360f1c1"}],"url":"https://sah.borca.ai/concepts/co_f571a333759802d5bd10d76ef8321665"}],"external_ids":{"DOI":"10.1109/TCSVT.2023.3239511","ArXiv":null,"PubMed":null,"PubMedCentral":null,"MAG":null,"DBLP":"journals/tcsv/JiangLZZZ23","ACL":null},"open_access":{"is_open_access":false,"pdf_url":null,"landing_url":"https://sah.borca.ai/papers/256291999","source":null,"pdf_url_source":null,"license":null,"reason":"pdf_url_not_indexed"},"reference_availability":{"status":"available","references_indexed":true,"full_text_available":false,"full_text_source":null,"count_basis":"semantic_scholar_metadata","extraction_status":"not_applicable","reason":null},"source":{"provider":"episteme2","base_corpus":"semantic_scholar_dump","freshness_mode":"unknown","basis":["semantic_scholar_metadata","postgres_metadata"],"limits":["paper metadata is based on indexed upstream scholarly datasets","claims and concepts are available only for extracted papers","absence of claims or concepts means no extracted graph data is available in this response"],"status":"available","degraded":false,"degraded_reasons":[],"diagnostics":{"status":"available","degraded":false,"degraded_reasons":[],"metadata_status":"available","graph_status":"available","abstract_status":"available"},"source_flags":1},"paper_id":631293,"paper_uid":"a8b856b1-ddc6-44ca-a41d-077e1e0aac0a","canonical_identity":{"paper_id":631293,"paper_uid":"a8b856b1-ddc6-44ca-a41d-077e1e0aac0a","identity_status":"available","lookup_basis":"semantic_scholar_external_id","compatibility_path":"corpus_id"},"url":"https://sah.borca.ai/papers/256291999"}