{"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/pyramid-mask-text-detector","title":"Pyramid Mask Text Detector","arxiv_id":"1903.11800","date":"2019-03-28","proceeding":null,"authors":["Jingchao Liu","Xuebo Liu","Jie Sheng","Ding Liang","Xin Li","Qingjie Liu"],"abstract":"Scene text detection, an essential step of scene text recognition system, is\nto locate text instances in natural scene images automatically. Some recent\nattempts benefiting from Mask R-CNN formulate scene text detection task as an\ninstance segmentation problem and achieve remarkable performance. In this\npaper, we present a new Mask R-CNN based framework named Pyramid Mask Text\nDetector (PMTD) to handle the scene text detection. Instead of binary text mask\ngenerated by the existing Mask R-CNN based methods, our PMTD performs\npixel-level regression under the guidance of location-aware supervision,\nyielding a more informative soft text mask for each text instance. As for the\ngeneration of text boxes, PMTD reinterprets the obtained 2D soft mask into 3D\nspace and introduces a novel plane clustering algorithm to derive the optimal\ntext box on the basis of 3D shape. Experiments on standard datasets demonstrate\nthat the proposed PMTD brings consistent and noticeable gain and clearly\noutperforms state-of-the-art methods. Specifically, it achieves an F-measure of\n80.13% on ICDAR 2017 MLT dataset.","url_abs":"http://arxiv.org/abs/1903.11800v1","url_pdf":"http://arxiv.org/pdf/1903.11800v1.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":"pyramid-mask-text-detector","repo_url":"https://github.com/anhnguyen9a7/pyramid-mask-and-plane-clustering-visualization","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null},{"paper_slug":"pyramid-mask-text-detector","repo_url":"https://github.com/jjprincess/PMTD","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"clustering","task_name":"Clustering"},{"task_slug":"instance-segmentation","task_name":"Instance Segmentation"},{"task_slug":"scene-text-detection","task_name":"Scene Text Detection"},{"task_slug":"scene-text-recognition","task_name":"Scene Text Recognition"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"},{"task_slug":"text-detection","task_name":"Text Detection"}],"methods":[{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"mask-r-cnn","method_name":"Mask R-CNN"},{"method_slug":"rpn","method_name":"RPN"},{"method_slug":"roi-align","method_name":"RoIAlign"},{"method_slug":"softmax","method_name":"Softmax"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/scene-text-detection-on-icdar-2015","task":"Scene Text Detection","dataset":"ICDAR 2015","model":"PMTD","rank_in_archive_order":10,"of":43,"metrics":{"F-Measure":"89.33","Precision":"91.3","Recall":"87.43"},"uses_additional_data":false},{"leaderboard":"/sota/scene-text-detection-on-icdar-2017-mlt-1","task":"Scene Text Detection","dataset":"ICDAR 2017 MLT","model":"PMTD*","rank_in_archive_order":1,"of":14,"metrics":{"F-Measure":"80.13%","Precision":"84.42","Recall":"76.25"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1903.11800","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}