Details
Presenter(s)
![Fanfu Xue Headshot](https://confcats-catavault.s3.amazonaws.com/CATAVault/ieeecass/master/files/styles/cc_user_photo/s3/user-pictures/12661_1.jpg?h=ad518777&itok=I1k03a76)
Display Name
Fanfu Xue
- Affiliation
-
AffiliationShandong Normal University
- Country
Abstract
In this paper, an end-to-end inpainting network for incomplete text (INIT), which can reconstruct the incomplete text into a complete text. On the incomplete text dataset reconstructed by INIT, the existing text recognition method can achieve high recognition accuracy, for instance ASTER can reach 82.99%, which verifies the effectiveness of our proposed INIT. Furthermore, we propose a method of incomplete text generation and build a synthesized incomplete text dataset.