{"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/small-scale-pedestrian-detection-based-on","title":"Small-scale Pedestrian Detection Based on Somatic Topology Localization and Temporal Feature Aggregation","arxiv_id":"1807.01438","date":"2018-07-04","proceeding":null,"authors":["Tao Song","Leiyu Sun","Di Xie","Haiming Sun","ShiLiang Pu"],"abstract":"A critical issue in pedestrian detection is to detect small-scale objects\nthat will introduce feeble contrast and motion blur in images and videos, which\nin our opinion should partially resort to deep-rooted annotation bias.\nMotivated by this, we propose a novel method integrated with somatic\ntopological line localization (TLL) and temporal feature aggregation for\ndetecting multi-scale pedestrians, which works particularly well with\nsmall-scale pedestrians that are relatively far from the camera. Moreover, a\npost-processing scheme based on Markov Random Field (MRF) is introduced to\neliminate ambiguities in occlusion cases. Applying with these methodologies\ncomprehensively, we achieve best detection performance on Caltech benchmark and\nimprove performance of small-scale objects significantly (miss rate decreases\nfrom 74.53% to 60.79%). Beyond this, we also achieve competitive performance on\nCityPersons dataset and show the existence of annotation bias in KITTI dataset.","url_abs":"http://arxiv.org/abs/1807.01438v1","url_pdf":"http://arxiv.org/pdf/1807.01438v1.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":"pedestrian-detection","task_name":"Pedestrian Detection"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/pedestrian-detection-on-citypersons","task":"Pedestrian Detection","dataset":"CityPersons","model":"TLL+MRF","rank_in_archive_order":18,"of":22,"metrics":{"Bare MR^-2":"9.2","Heavy MR^-2":"52.0","Partial MR^-2":"15.9","Reasonable MR^-2":"14.4"},"uses_additional_data":false},{"leaderboard":"/sota/pedestrian-detection-on-citypersons","task":"Pedestrian Detection","dataset":"CityPersons","model":"TLL","rank_in_archive_order":21,"of":22,"metrics":{"Bare MR^-2":"10.0","Heavy MR^-2":"53.6","Partial MR^-2":"17.2","Reasonable MR^-2":"15.5"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}