{"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/understanding-traffic-density-from-large","title":"Understanding Traffic Density from Large-Scale Web Camera Data","arxiv_id":"1703.05868","date":"2017-03-17","proceeding":"CVPR 2017 7","authors":["Shanghang Zhang","Guanhang Wu","João P. Costeira","José M. F. Moura"],"abstract":"Understanding traffic density from large-scale web camera (webcam) videos is\na challenging problem because such videos have low spatial and temporal\nresolution, high occlusion and large perspective. To deeply understand traffic\ndensity, we explore both deep learning based and optimization based methods. To\navoid individual vehicle detection and tracking, both methods map the image\ninto vehicle density map, one based on rank constrained regression and the\nother one based on fully convolution networks (FCN). The regression based\nmethod learns different weights for different blocks in the image to increase\nfreedom degrees of weights and embed perspective information. The FCN based\nmethod jointly estimates vehicle density map and vehicle count with a residual\nlearning framework to perform end-to-end dense prediction, allowing arbitrary\nimage resolution, and adapting to different vehicle scales and perspectives. We\nanalyze and compare both methods, and get insights from optimization based\nmethod to improve deep model. Since existing datasets do not cover all the\nchallenges in our work, we collected and labelled a large-scale traffic video\ndataset, containing 60 million frames from 212 webcams. Both methods are\nextensively evaluated and compared on different counting tasks and datasets.\nFCN based method significantly reduces the mean absolute error from 10.99 to\n5.31 on the public dataset TRANCOS compared with the state-of-the-art baseline.","url_abs":"http://arxiv.org/abs/1703.05868v3","url_pdf":"http://arxiv.org/pdf/1703.05868v3.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":"understanding-traffic-density-from-large","repo_url":"https://github.com/polltooh/traffic_video_analysis","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"regression-1","task_name":"regression"},{"task_slug":"vehicle-detection","task_name":"vehicle detection"}],"methods":[{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"fcn","method_name":"FCN"},{"method_slug":"max-pooling","method_name":"Max Pooling"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1703.05868","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}