{"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/derpn-taking-a-further-step-toward-more","title":"DeRPN: Taking a further step toward more general object detection","arxiv_id":"1811.06700","date":"2018-11-16","proceeding":null,"authors":["Lele Xie","Yuliang Liu","Lianwen Jin","Zecheng Xie"],"abstract":"Most current detection methods have adopted anchor boxes as regression\nreferences. However, the detection performance is sensitive to the setting of\nthe anchor boxes. A proper setting of anchor boxes may vary significantly\nacross different datasets, which severely limits the universality of the\ndetectors. To improve the adaptivity of the detectors, in this paper, we\npresent a novel dimension-decomposition region proposal network (DeRPN) that\ncan perfectly displace the traditional Region Proposal Network (RPN). DeRPN\nutilizes an anchor string mechanism to independently match object widths and\nheights, which is conducive to treating variant object shapes. In addition, a\nnovel scale-sensitive loss is designed to address the imbalanced loss\ncomputations of different scaled objects, which can avoid the small objects\nbeing overwhelmed by larger ones. Comprehensive experiments conducted on both\ngeneral object detection datasets (Pascal VOC 2007, 2012 and MS COCO) and scene\ntext detection datasets (ICDAR 2013 and COCO-Text) all prove that our DeRPN can\nsignificantly outperform RPN. It is worth mentioning that the proposed DeRPN\ncan be employed directly on different models, tasks, and datasets without any\nmodifications of hyperparameters or specialized optimization, which further\ndemonstrates its adaptivity. The code will be released at\nhttps://github.com/HCIILAB/DeRPN.","url_abs":"http://arxiv.org/abs/1811.06700v1","url_pdf":"http://arxiv.org/pdf/1811.06700v1.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":"derpn-taking-a-further-step-toward-more","repo_url":"https://github.com/HCIILAB/DeRPN","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"object","task_name":"Object"},{"task_slug":"object-detection","task_name":"Object Detection"},{"task_slug":"region-proposal","task_name":"Region Proposal"},{"task_slug":"scene-text-detection","task_name":"Scene Text Detection"},{"task_slug":"text-detection","task_name":"Text Detection"},{"task_slug":"object-detection-1","task_name":"object-detection"}],"methods":[{"method_slug":"rpn","method_name":"RPN"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}