{"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/dgc-net-dense-geometric-correspondence","title":"DGC-Net: Dense Geometric Correspondence Network","arxiv_id":"1810.08393","date":"2018-10-19","proceeding":null,"authors":["Iaroslav Melekhov","Aleksei Tiulpin","Torsten Sattler","Marc Pollefeys","Esa Rahtu","Juho Kannala"],"abstract":"This paper addresses the challenge of dense pixel correspondence estimation\nbetween two images. This problem is closely related to optical flow estimation\ntask where ConvNets (CNNs) have recently achieved significant progress. While\noptical flow methods produce very accurate results for the small pixel\ntranslation and limited appearance variation scenarios, they hardly deal with\nthe strong geometric transformations that we consider in this work. In this\npaper, we propose a coarse-to-fine CNN-based framework that can leverage the\nadvantages of optical flow approaches and extend them to the case of large\ntransformations providing dense and subpixel accurate estimates. It is trained\non synthetic transformations and demonstrates very good performance to unseen,\nrealistic, data. Further, we apply our method to the problem of relative camera\npose estimation and demonstrate that the model outperforms existing dense\napproaches.","url_abs":"http://arxiv.org/abs/1810.08393v2","url_pdf":"http://arxiv.org/pdf/1810.08393v2.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":"dgc-net-dense-geometric-correspondence","repo_url":"https://github.com/AaltoVision/DGC-Net","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"NOASSERTION"}},{"paper_slug":"dgc-net-dense-geometric-correspondence","repo_url":"https://github.com/2023-MindSpore-1/ms-code-211/tree/main/dgcnet_res101","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null},{"paper_slug":"dgc-net-dense-geometric-correspondence","repo_url":"https://github.com/MindSpore-paper-code-3/code8/tree/main/dgcnet_res101","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null},{"paper_slug":"dgc-net-dense-geometric-correspondence","repo_url":"https://github.com/code-implementation1/Code1/tree/main/dgcnet_res101","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null}],"tasks":[{"task_slug":"camera-pose-estimation","task_name":"Camera Pose Estimation"},{"task_slug":"dense-pixel-correspondence-estimation","task_name":"Dense Pixel Correspondence Estimation"},{"task_slug":"optical-flow-estimation","task_name":"Optical Flow Estimation"},{"task_slug":"translation","task_name":"Translation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/dense-pixel-correspondence-estimation-on","task":"Dense Pixel Correspondence Estimation","dataset":"HPatches","model":"DGC-Net aff+tps+homo","rank_in_archive_order":2,"of":8,"metrics":{"Viewpoint I AEPE":"1.55","Viewpoint II AEPE":"5.53","Viewpoint III AEPE":"8.98","Viewpoint IV AEPE":"11.66","Viewpoint V AEPE":"16.70"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1810.08393","atlas_url":"https://app.syntology.ai/?focus=1810.08393","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}