{"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/dense-depth-posterior-ddp-from-single-image","title":"Dense Depth Posterior (DDP) from Single Image and Sparse Range","arxiv_id":"1901.10034","date":"2019-01-28","proceeding":"CVPR 2019 6","authors":["Yanchao Yang","Alex Wong","Stefano Soatto"],"abstract":"We present a deep learning system to infer the posterior distribution of a\ndense depth map associated with an image, by exploiting sparse range\nmeasurements, for instance from a lidar. While the lidar may provide a depth\nvalue for a small percentage of the pixels, we exploit regularities reflected\nin the training set to complete the map so as to have a probability over depth\nfor each pixel in the image. We exploit a Conditional Prior Network, that\nallows associating a probability to each depth value given an image, and\ncombine it with a likelihood term that uses the sparse measurements. Optionally\nwe can also exploit the availability of stereo during training, but in any case\nonly require a single image and a sparse point cloud at run-time. We test our\napproach on both unsupervised and supervised depth completion using the KITTI\nbenchmark, and improve the state-of-the-art in both.","url_abs":"http://arxiv.org/abs/1901.10034v2","url_pdf":"http://arxiv.org/pdf/1901.10034v2.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":[],"tasks":[{"task_slug":"depth-completion","task_name":"Depth Completion"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/depth-completion-on-void","task":"Depth Completion","dataset":"VOID","model":"DDP","rank_in_archive_order":5,"of":6,"metrics":{"MAE":"151.86","RMSE":"222.36","iMAE":"74.59","iRMSE":"112.36"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1901.10034","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}