{"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/hydraplus-net-attentive-deep-features-for","title":"HydraPlus-Net: Attentive Deep Features for Pedestrian Analysis","arxiv_id":"1709.09930","date":"2017-09-28","proceeding":"ICCV 2017 10","authors":["Xihui Liu","Haiyu Zhao","Maoqing Tian","Lu Sheng","Jing Shao","Shuai Yi","Junjie Yan","Xiaogang Wang"],"abstract":"Pedestrian analysis plays a vital role in intelligent video surveillance and\nis a key component for security-centric computer vision systems. Despite that\nthe convolutional neural networks are remarkable in learning discriminative\nfeatures from images, the learning of comprehensive features of pedestrians for\nfine-grained tasks remains an open problem. In this study, we propose a new\nattention-based deep neural network, named as HydraPlus-Net (HP-net), that\nmulti-directionally feeds the multi-level attention maps to different feature\nlayers. The attentive deep features learned from the proposed HP-net bring\nunique advantages: (1) the model is capable of capturing multiple attentions\nfrom low-level to semantic-level, and (2) it explores the multi-scale\nselectiveness of attentive features to enrich the final feature representations\nfor a pedestrian image. We demonstrate the effectiveness and generality of the\nproposed HP-net for pedestrian analysis on two tasks, i.e. pedestrian attribute\nrecognition and person re-identification. Intensive experimental results have\nbeen provided to prove that the HP-net outperforms the state-of-the-art methods\non various datasets.","url_abs":"http://arxiv.org/abs/1709.09930v1","url_pdf":"http://arxiv.org/pdf/1709.09930v1.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":"hydraplus-net-attentive-deep-features-for","repo_url":"https://github.com/xh-liu/HydraPlus-Net","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok"}},{"paper_slug":"hydraplus-net-attentive-deep-features-for","repo_url":"https://github.com/TianmingQiu/HydraPlusNet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"attribute","task_name":"Attribute"},{"task_slug":"pedestrian-attribute-recognition","task_name":"Pedestrian Attribute Recognition"},{"task_slug":"person-re-identification","task_name":"Person Re-Identification"}],"methods":[],"datasets_introduced":[{"slug":"pa-100k","name":"PA-100K","full_name":"PA-100K Dataset"}],"methods_introduced":[],"results":[{"leaderboard":"/sota/pedestrian-attribute-recognition-on-pa-100k","task":"Pedestrian Attribute Recognition","dataset":"PA-100K","model":"HP-net","rank_in_archive_order":11,"of":13,"metrics":{"Accuracy":"72.19%"},"uses_additional_data":false},{"leaderboard":"/sota/pedestrian-attribute-recognition-on-peta","task":"Pedestrian Attribute Recognition","dataset":"PETA","model":"HP-net","rank_in_archive_order":6,"of":6,"metrics":{"Accuracy":"76.13%"},"uses_additional_data":false},{"leaderboard":"/sota/pedestrian-attribute-recognition-on-rap","task":"Pedestrian Attribute Recognition","dataset":"RAP","model":"HP-net","rank_in_archive_order":2,"of":2,"metrics":{"Accuracy":"65.39%"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1709.09930","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}