{"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/learning-intrinsic-image-decomposition-from","title":"Learning Intrinsic Image Decomposition from Watching the World","arxiv_id":"1804.00582","date":"2018-04-02","proceeding":"CVPR 2018 6","authors":["Zhengqi Li","Noah Snavely"],"abstract":"Single-view intrinsic image decomposition is a highly ill-posed problem, and\nso a promising approach is to learn from large amounts of data. However, it is\ndifficult to collect ground truth training data at scale for intrinsic images.\nIn this paper, we explore a different approach to learning intrinsic images:\nobserving image sequences over time depicting the same scene under changing\nillumination, and learning single-view decompositions that are consistent with\nthese changes. This approach allows us to learn without ground truth\ndecompositions, and to instead exploit information available from multiple\nimages when training. Our trained model can then be applied at test time to\nsingle views. We describe a new learning framework based on this idea,\nincluding new loss functions that can be efficiently evaluated over entire\nsequences. While prior learning-based methods achieve good performance on\nspecific benchmarks, we show that our approach generalizes well to several\ndiverse datasets, including MIT intrinsic images, Intrinsic Images in the Wild\nand Shading Annotations in the Wild.","url_abs":"http://arxiv.org/abs/1804.00582v1","url_pdf":"http://arxiv.org/pdf/1804.00582v1.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":"learning-intrinsic-image-decomposition-from","repo_url":"https://github.com/lixx2938/unsupervised-learning-intrinsic-images","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"intrinsic-image-decomposition","task_name":"Intrinsic Image Decomposition"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1804.00582","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}