{"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/lego-learning-edge-with-geometry-all-at-once","title":"LEGO: Learning Edge with Geometry all at Once by Watching Videos","arxiv_id":"1803.05648","date":"2018-03-15","proceeding":"CVPR 2018 6","authors":["Zhenheng Yang","Peng Wang","Yang Wang","Wei Xu","Ram Nevatia"],"abstract":"Learning to estimate 3D geometry in a single image by watching unlabeled\nvideos via deep convolutional network is attracting significant attention. In\nthis paper, we introduce a \"3D as-smooth-as-possible (3D-ASAP)\" prior inside\nthe pipeline, which enables joint estimation of edges and 3D scene, yielding\nresults with significant improvement in accuracy for fine detailed structures.\nSpecifically, we define the 3D-ASAP prior by requiring that any two points\nrecovered in 3D from an image should lie on an existing planar surface if no\nother cues provided. We design an unsupervised framework that Learns Edges and\nGeometry (depth, normal) all at Once (LEGO). The predicted edges are embedded\ninto depth and surface normal smoothness terms, where pixels without edges\nin-between are constrained to satisfy the prior. In our framework, the\npredicted depths, normals and edges are forced to be consistent all the time.\nWe conduct experiments on KITTI to evaluate our estimated geometry and\nCityScapes to perform edge evaluation. We show that in all of the tasks,\ni.e.depth, normal and edge, our algorithm vastly outperforms other\nstate-of-the-art (SOTA) algorithms, demonstrating the benefits of our approach.","url_abs":"http://arxiv.org/abs/1803.05648v2","url_pdf":"http://arxiv.org/pdf/1803.05648v2.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":"lego-learning-edge-with-geometry-all-at-once","repo_url":"https://github.com/zhenheny/LEGO","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"3d-geometry","task_name":"3D geometry"},{"task_slug":"all","task_name":"All"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1803.05648","atlas_url":"https://app.syntology.ai/?focus=1803.05648","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}