{"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/deep-joint-rain-detection-and-removal-from-a","title":"Deep Joint Rain Detection and Removal from a Single Image","arxiv_id":"1609.07769","date":"2016-09-25","proceeding":"CVPR 2017 7","authors":["Wenhan Yang","Robby T. Tan","Jiashi Feng","Jiaying Liu","Zongming Guo","Shuicheng Yan"],"abstract":"In this paper, we address a rain removal problem from a single image, even in\nthe presence of heavy rain and rain streak accumulation. Our core ideas lie in\nthe new rain image models and a novel deep learning architecture. We first\nmodify an existing model comprising a rain streak layer and a background layer,\nby adding a binary map that locates rain streak regions. Second, we create a\nnew model consisting of a component representing rain streak accumulation\n(where individual streaks cannot be seen, and thus visually similar to mist or\nfog), and another component representing various shapes and directions of\noverlapping rain streaks, which usually happen in heavy rain. Based on the\nfirst model, we develop a multi-task deep learning architecture that learns the\nbinary rain streak map, the appearance of rain streaks, and the clean\nbackground, which is our ultimate output. The additional binary map is\ncritically beneficial, since its loss function can provide additional strong\ninformation to the network. To handle rain streak accumulation (again, a\nphenomenon visually similar to mist or fog) and various shapes and directions\nof overlapping rain streaks, we propose a recurrent rain detection and removal\nnetwork that removes rain streaks and clears up the rain accumulation\niteratively and progressively. In each recurrence of our method, a new\ncontextualized dilated network is developed to exploit regional contextual\ninformation and outputs better representation for rain detection. The\nevaluation on real images, particularly on heavy rain, shows the effectiveness\nof our novel models and architecture, outperforming the state-of-the-art\nmethods significantly. Our codes and data sets will be publicly available.","url_abs":"http://arxiv.org/abs/1609.07769v3","url_pdf":"http://arxiv.org/pdf/1609.07769v3.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":"deep-joint-rain-detection-and-removal-from-a","repo_url":"https://github.com/ZhangXinNan/RainDetectionAndRemoval","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}},{"paper_slug":"deep-joint-rain-detection-and-removal-from-a","repo_url":"https://github.com/jiupinjia/deep-adversarial-decomposition","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"rain-removal","task_name":"Rain Removal"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1609.07769","atlas_url":"https://app.syntology.ai/?focus=1609.07769","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}