Scene text removal via cascaded text stroke detection and erasing

作     者：Xuewei Bian Chaoqun Wang Weize Quan Juntao Ye Xiaopeng Zhang Dong-Ming Yan Xuewei Bian;Chaoqun Wang;Weize Quan;Juntao Ye;Xiaopeng Zhang;Dong-Ming Yan

作者机构：National Laboratory of Pattern RecognitionInstitute of AutomationBeijing 100049China School of Artificial IntelligenceUniversity of Chinese Academy of SciencesBeijing 100084China 

出 版 物：《Computational Visual Media》 (计算可视媒体（英文版）)

年 卷 期：2022年第8卷第2期

页      面：273-287页

核心收录：

学科分类：08[工学] 080203[工学-机械设计及理论] 0802[工学-机械工程] 

基　　金：supported by the National Natural Science Foundation of China(62102418 and 62172415) the National Key R&D Program of China(2019YFB2204104) the Open Research Fund Program of State key Laboratory of Hydroscience and Engineering,Tsinghua University(sklhse-2020-D-07) 

主　　题：scene text removal text stroke detection generative adversarial networks cascaded network design real-world dataset 

摘      要：Recent learning-based approaches show promising performance improvement for the scene text removal task but usually leave several remnants of text and provide visually unpleasant *** this work,a novel end-to-end framework is proposed based on accurate text stroke ***,the text removal problem is decoupled into text stroke detection and stroke removal;we design separate networks to solve these two subproblems,the latter being a generative *** two networks are combined as a processing unit,which is cascaded to obtain our final model for text *** results demonstrate that the proposed method substantially outperforms the state-of-the-art for locating and erasing scene text.A new large-scale real-world dataset with 12,120 images has been constructed and is being made available to facilitate research,as current publicly available datasets are mainly synthetic so cannot properly measure the performance of different methods.