{"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/wespe-weakly-supervised-photo-enhancer-for","title":"WESPE: Weakly Supervised Photo Enhancer for Digital Cameras","arxiv_id":"1709.01118","date":"2017-09-04","proceeding":null,"authors":["Andrey Ignatov","Nikolay Kobyshev","Radu Timofte","Kenneth Vanhoey","Luc van Gool"],"abstract":"Low-end and compact mobile cameras demonstrate limited photo quality mainly\ndue to space, hardware and budget constraints. In this work, we propose a deep\nlearning solution that translates photos taken by cameras with limited\ncapabilities into DSLR-quality photos automatically. We tackle this problem by\nintroducing a weakly supervised photo enhancer (WESPE) - a novel image-to-image\nGenerative Adversarial Network-based architecture. The proposed model is\ntrained by under weak supervision: unlike previous works, there is no need for\nstrong supervision in the form of a large annotated dataset of aligned\noriginal/enhanced photo pairs. The sole requirement is two distinct datasets:\none from the source camera, and one composed of arbitrary high-quality images\nthat can be generally crawled from the Internet - the visual content they\nexhibit may be unrelated. Hence, our solution is repeatable for any camera:\ncollecting the data and training can be achieved in a couple of hours. In this\nwork, we emphasize on extensive evaluation of obtained results. Besides\nstandard objective metrics and subjective user study, we train a virtual rater\nin the form of a separate CNN that mimics human raters on Flickr data and use\nthis network to get reference scores for both original and enhanced photos. Our\nexperiments on the DPED, KITTI and Cityscapes datasets as well as pictures from\nseveral generations of smartphones demonstrate that WESPE produces comparable\nor improved qualitative results with state-of-the-art strongly supervised\nmethods.","url_abs":"http://arxiv.org/abs/1709.01118v2","url_pdf":"http://arxiv.org/pdf/1709.01118v2.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":"wespe-weakly-supervised-photo-enhancer-for","repo_url":"https://github.com/GBATZOLIS/Wespe-Keras","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}},{"paper_slug":"wespe-weakly-supervised-photo-enhancer-for","repo_url":"https://github.com/kirkutirev/photo_enhancer","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"wespe-weakly-supervised-photo-enhancer-for","repo_url":"https://github.com/sanityseeker/wespe","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":null,"task_name":"Generative Adversarial Network"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1709.01118","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}