{"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/tightness-aware-evaluation-protocol-for-scene","title":"Tightness-aware Evaluation Protocol for Scene Text Detection","arxiv_id":"1904.00813","date":"2019-03-27","proceeding":"CVPR 2019 6","authors":["Yuliang Liu","Lianwen Jin","Zecheng Xie","Canjie Luo","Shuaitao Zhang","Lele Xie"],"abstract":"Evaluation protocols play key role in the developmental progress of text\ndetection methods. There are strict requirements to ensure that the evaluation\nmethods are fair, objective and reasonable. However, existing metrics exhibit\nsome obvious drawbacks: 1) They are not goal-oriented; 2) they cannot recognize\nthe tightness of detection methods; 3) existing one-to-many and many-to-one\nsolutions involve inherent loopholes and deficiencies. Therefore, this paper\nproposes a novel evaluation protocol called Tightness-aware\nIntersect-over-Union (TIoU) metric that could quantify completeness of ground\ntruth, compactness of detection, and tightness of matching degree.\nSpecifically, instead of merely using the IoU value, two common detection\nbehaviors are properly considered; meanwhile, directly using the score of TIoU\nto recognize the tightness. In addition, we further propose a straightforward\nmethod to address the annotation granularity issue, which can fairly evaluate\nword and text-line detections simultaneously. By adopting the detection results\nfrom published methods and general object detection frameworks, comprehensive\nexperiments on ICDAR 2013 and ICDAR 2015 datasets are conducted to compare\nrecent metrics and the proposed TIoU metric. The comparison demonstrated some\npromising new prospects, e.g., determining the methods and frameworks for which\nthe detection is tighter and more beneficial to recognize. Our method is\nextremely simple; however, the novelty is none other than the proposed metric\ncan utilize simplest but reasonable improvements to lead to many interesting\nand insightful prospects and solving most the issues of the previous metrics.\nThe code is publicly available at https://github.com/Yuliang-Liu/TIoU-metric .","url_abs":"http://arxiv.org/abs/1904.00813v1","url_pdf":"http://arxiv.org/pdf/1904.00813v1.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":"tightness-aware-evaluation-protocol-for-scene","repo_url":"https://github.com/Yuliang-Liu/TIoU-metric","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"object-detection","task_name":"Object Detection"},{"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":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1904.00813","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}