{"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/adcrowdnet-an-attention-injective-deformable","title":"ADCrowdNet: An Attention-injective Deformable Convolutional Network for Crowd Understanding","arxiv_id":"1811.11968","date":"2018-11-29","proceeding":"CVPR 2019 6","authors":["Ning Liu","Yongchao Long","Changqing Zou","Qun Niu","Li Pan","Hefeng Wu"],"abstract":"We propose an attention-injective deformable convolutional network called\nADCrowdNet for crowd understanding that can address the accuracy degradation\nproblem of highly congested noisy scenes. ADCrowdNet contains two concatenated\nnetworks. An attention-aware network called Attention Map Generator (AMG) first\ndetects crowd regions in images and computes the congestion degree of these\nregions. Based on detected crowd regions and congestion priors, a multi-scale\ndeformable network called Density Map Estimator (DME) then generates\nhigh-quality density maps. With the attention-aware training scheme and\nmulti-scale deformable convolutional scheme, the proposed ADCrowdNet achieves\nthe capability of being more effective to capture the crowd features and more\nresistant to various noises. We have evaluated our method on four popular crowd\ncounting datasets (ShanghaiTech, UCF_CC_50, WorldEXPO'10, and UCSD) and an\nextra vehicle counting dataset TRANCOS, and our approach beats existing\nstate-of-the-art approaches on all of these datasets.","url_abs":"http://arxiv.org/abs/1811.11968v5","url_pdf":"http://arxiv.org/pdf/1811.11968v5.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":"adcrowdnet-an-attention-injective-deformable","repo_url":"https://github.com/BIGKnight/ADCrowd_pytorch_implementation","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"crowd-counting","task_name":"Crowd Counting"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/crowd-counting-on-trancos","task":"Crowd Counting","dataset":"TRANCOS","model":"ADCrowdNet","rank_in_archive_order":2,"of":4,"metrics":{"MAE":"2.44"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1811.11968","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}