{"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/cascade-residual-learning-a-two-stage","title":"Cascade Residual Learning: A Two-stage Convolutional Neural Network for Stereo Matching","arxiv_id":"1708.09204","date":"2017-08-30","proceeding":null,"authors":["Jiahao Pang","Wenxiu Sun","Jimmy SJ. Ren","Chengxi Yang","Qiong Yan"],"abstract":"Leveraging on the recent developments in convolutional neural networks\n(CNNs), matching dense correspondence from a stereo pair has been cast as a\nlearning problem, with performance exceeding traditional approaches. However,\nit remains challenging to generate high-quality disparities for the inherently\nill-posed regions. To tackle this problem, we propose a novel cascade CNN\narchitecture composing of two stages. The first stage advances the recently\nproposed DispNet by equipping it with extra up-convolution modules, leading to\ndisparity images with more details. The second stage explicitly rectifies the\ndisparity initialized by the first stage; it couples with the first-stage and\ngenerates residual signals across multiple scales. The summation of the outputs\nfrom the two stages gives the final disparity. As opposed to directly learning\nthe disparity at the second stage, we show that residual learning provides more\neffective refinement. Moreover, it also benefits the training of the overall\ncascade network. Experimentation shows that our cascade residual learning\nscheme provides state-of-the-art performance for matching stereo\ncorrespondence. By the time of the submission of this paper, our method ranks\nfirst in the KITTI 2015 stereo benchmark, surpassing the prior works by a\nnoteworthy margin.","url_abs":"http://arxiv.org/abs/1708.09204v2","url_pdf":"http://arxiv.org/pdf/1708.09204v2.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":"cascade-residual-learning-a-two-stage","repo_url":"https://github.com/Artifineuro/crl","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"stereo-matching-1","task_name":"Stereo Matching"},{"task_slug":"stereo-matching","task_name":"Stereo Matching Hand"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1708.09204","atlas_url":"https://app.syntology.ai/?focus=1708.09204","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}