STAN: A sequential transformation attention-based network for scene text recognition

作者：

Highlights：

• We propose a new method called sequential transformation attention network for irregular scene text recognition.

• Taking decomposition as the key idea, our method can rectify irregular text effectively.

• The proposed method is optimized in an end-to-end weakly supervised manner.

• Our method achieves state-of-the-art results on text recognition benchmarks.

摘要

•We propose a new method called sequential transformation attention network for irregular scene text recognition.•Taking decomposition as the key idea, our method can rectify irregular text effectively.•The proposed method is optimized in an end-to-end weakly supervised manner.•Our method achieves state-of-the-art results on text recognition benchmarks.

论文关键词：Scene text recognition,Scene text rectification,Optical character recognition,Deep learning

论文评审过程：Received 30 March 2019, Revised 6 October 2020, Accepted 7 October 2020, Available online 9 October 2020, Version of Record 16 October 2020.

论文官网地址：https://doi.org/10.1016/j.patcog.2020.107692