CCANet: Exploiting Pixel-wise Semantics for Irregular Scene Text Spotting | |
Xu, Shanbo1,2; Chen, Chen1,2; Peng, Silong1,2; Hu, Xiyuan3 | |
2021-10 | |
会议日期 | 2021-10 |
会议地点 | Shanghai, China |
关键词 | computer vision scene text spotting irregular text |
DOI | 10.1109/CISP-BMEI53629.2021.9624403 |
英文摘要 | Despite the progress in regular scene text spotting, how to detect and recognize irregular text with efficiency and accuracy remains a challenging task. In this work, we propose a novel Corner and Character Assisted Network (CCANet) which exploits pixel-wise semantics to learn explicit text corner and character center positions with low computational cost. Concretely, in the detection stage, we develop a pixel-level Corner Rectification Branch to refine the inaccurately regressed text corners; in the recognition stage, we design another pixellevel Character Enhancement Branch which generates a Gaussian-like character center heatmap to provide attention guidance for the decoding process. To overcome the reliance of character-level annotations, we adopt an iterative approach to generate pseudo-GT label for the character heatmap, which regards the attention peak position of the attention-based recognizer as the true character center. The extensive experiments conducted on two irregular text benchmarks, TotalText and CTW1500, demonstrate that the proposed CCANet achieves competitive and even new state-of-the-art performance. |
会议录 | 2021 14th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI) |
会议录出版者 | IEEE |
语种 | 英语 |
内容类型 | 会议论文 |
源URL | [http://ir.ia.ac.cn/handle/173211/48529] |
专题 | 自动化研究所_智能制造技术与系统研究中心_多维数据分析团队 |
通讯作者 | Chen, Chen |
作者单位 | 1.University of Chinese Academy of Sciences 2.Institute of Automation, Chinese Academy of Sciences 3.Nanjing University of Science and Technology |
推荐引用方式 GB/T 7714 | Xu, Shanbo,Chen, Chen,Peng, Silong,et al. CCANet: Exploiting Pixel-wise Semantics for Irregular Scene Text Spotting[C]. 见:. Shanghai, China. 2021-10. |
个性服务 |
查看访问统计 |
相关权益政策 |
暂无数据 |
收藏/分享 |
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。
修改评论