{"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/repulsion-loss-detecting-pedestrians-in-a","title":"Repulsion Loss: Detecting Pedestrians in a Crowd","arxiv_id":"1711.07752","date":"2017-11-21","proceeding":"CVPR 2018 6","authors":["Xinlong Wang","Tete Xiao","Yuning Jiang","Shuai Shao","Jian Sun","Chunhua Shen"],"abstract":"Detecting individual pedestrians in a crowd remains a challenging problem\nsince the pedestrians often gather together and occlude each other in\nreal-world scenarios. In this paper, we first explore how a state-of-the-art\npedestrian detector is harmed by crowd occlusion via experimentation, providing\ninsights into the crowd occlusion problem. Then, we propose a novel bounding\nbox regression loss specifically designed for crowd scenes, termed repulsion\nloss. This loss is driven by two motivations: the attraction by target, and the\nrepulsion by other surrounding objects. The repulsion term prevents the\nproposal from shifting to surrounding objects thus leading to more crowd-robust\nlocalization. Our detector trained by repulsion loss outperforms all the\nstate-of-the-art methods with a significant improvement in occlusion cases.","url_abs":"http://arxiv.org/abs/1711.07752v2","url_pdf":"http://arxiv.org/pdf/1711.07752v2.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":"repulsion-loss-detecting-pedestrians-in-a","repo_url":"https://github.com/bailvwangzi/repulsion_loss_ssd","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"repulsion-loss-detecting-pedestrians-in-a","repo_url":"https://github.com/justinkay/repulsion-loss-detectron2","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"pedestrian-detection","task_name":"Pedestrian Detection"},{"task_slug":"regression-1","task_name":"regression"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/pedestrian-detection-on-caltech","task":"Pedestrian Detection","dataset":"Caltech","model":"RepLoss + CityPersons dataset","rank_in_archive_order":9,"of":33,"metrics":{"Reasonable Miss Rate":"4.0"},"uses_additional_data":true},{"leaderboard":"/sota/pedestrian-detection-on-caltech","task":"Pedestrian Detection","dataset":"Caltech","model":"RepLoss","rank_in_archive_order":13,"of":33,"metrics":{"Reasonable Miss Rate":"5.0"},"uses_additional_data":false},{"leaderboard":"/sota/pedestrian-detection-on-citypersons","task":"Pedestrian Detection","dataset":"CityPersons","model":"RepLoss","rank_in_archive_order":17,"of":22,"metrics":{"Bare MR^-2":"7.6","Heavy MR^-2":"56.9","Partial MR^-2":"16.8","Reasonable MR^-2":"13.2"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1711.07752","atlas_url":"https://app.syntology.ai/?focus=1711.07752","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}